CN114830255A - Information transmission device and information transmission method - Google Patents

Abstract

The invention provides an information transmission device and an information transmission method, which can grasp accurate health status by considering the status of a target person and provide customized information such as suggestions corresponding to the health status. Comprising: an ID determination unit (1b) which acquires a 1 st inspection data set of a subject person in time series by a 1 st device (2a) and acquires a 2 nd inspection data set of the subject person in time series by a 2 nd device (2b), wherein the 2 nd device (2b) can perform an inspection capable of complementing the 1 st inspection data set; and an information providing unit (1c) for determining delivery information to be provided to the subject person by using the 1 st inspection data group and the 2 nd inspection data group, the 1 st inspection data group and the 2 nd inspection data group complementing each other with inspection time or inspection items.

Description

Information transmission device and information transmission method

Technical Field

The present invention relates to an information delivery apparatus and an information delivery method capable of providing customized information such as advice to a user based on an examination result that can be obtained in daily life.

Background

In recent years, the internet has become widespread, and information closely related to the life of a user can be easily acquired by using the internet. Services for providing various customized information (effective information) corresponding to each user by using the acquired information are increasing. For example, a service introducing health foods and the like is information that attracts common interest to many people, and thus, many of such services can be seen.

In addition, since the network environment has been improved, various proposals have been made for remote examination outside professional institutions such as hospitals. For example, patent document 1 discloses a remote inspection method in which a sensor chip and a mobile phone are used as a reader/writer and inspection data is transmitted by using a common communication network. Patent document 1 proposes a method of storing past inspection data and evaluation results thereof in a database and using the information.

Further, patent document 2 discloses a biological information measurement device that integrates personal authentication data and image data captured by an excrement capture means and transmits the integrated data via a communication means. Patent document 3 discloses a method for displaying medical information, the method including: and a history browser for displaying a learning list related to the examination, displaying a medical image based on the medical image information of the examinee in the selected learning on the image display screen, and displaying the learning list representing the examinee when the history browser is requested.

Documents of the prior art

Patent document

Patent document 1: japanese laid-open patent publication No. 2009-258886

Patent document 2: japanese patent laid-open No. 2014-031655

Patent document 3: japanese patent No. 5294947

Disclosure of Invention

Problems to be solved by the invention

Patent documents 1 to 3 describe that biometric information is acquired and transmitted remotely to use the content of the information. In the case where the subject does not have subjective symptoms, it is advantageous to be able to know whether the subject has a disease or may have a disease. Not only a plurality of hospitals and examination facilities, users may receive examinations using examination equipment equipped at their homes, offices, and the like. With regard to the inspection equipment, although there is a possibility that various devices may be used even if the inspection items are the same, this is not considered in the above-mentioned patent documents 1 to 3.

The present invention has been made in view of the above circumstances, and an object thereof is to provide an information delivery apparatus and an information delivery method capable of grasping an accurate health state in consideration of the condition of a subject person and providing customized information such as advice corresponding to the health state.

Means for solving the problems

In order to achieve the above object, the information delivery apparatus of claim 1 includes: a 1 st inspection data acquisition unit that acquires a 1 st inspection data set of a subject person as a time series by a 1 st device; and a 2 nd inspection data acquisition unit that acquires a 2 nd inspection data set of the subject person as a time series by a 2 nd device, the 2 nd device being capable of performing an inspection capable of complementing the 1 st inspection data set; and a delivery information determining unit that determines delivery information to be provided to the subject using the 1 st examination data group and the 2 nd examination data group, the 1 st examination data group and the 2 nd examination data group complementing each other with an examination time or an examination item.

The information delivery apparatus according to claim 2 is the information delivery apparatus according to claim 1, wherein the delivery information determining unit determines the delivery information according to an inference model learned according to a change pattern of the inspection data set acquired by the plurality of devices.

In the information delivery apparatus according to claim 3 of the present invention as set forth in claim 1, the delivery information determining unit corrects the 1 st inspection data group and the 2 nd inspection data group in accordance with the 1 st inspection data group acquired by the 1 st device and the 2 nd inspection data group acquired by the 2 nd device, calculates the reliability when the inference is made using the corrected inspection data group as an input, and determines the delivery information in accordance with the reliability.

In the information transmission device according to claim 4, in the above-mentioned 3, the transmission information determining unit performs a four-way operation on a numerical value common to the respective data included in the inspection data group when performing the correction for each of the inspection data groups.

In the information delivery apparatus according to claim 5 of the invention described in claim 1, the delivery information determining unit corrects the 1 st inspection data group and the 2 nd inspection data group in accordance with the 1 st inspection data group acquired by the 1 st device and the 2 nd inspection data group acquired by the 2 nd device, combines the corrected inspection data groups into 1 inspection data group, inputs the combined inspection data group to an inference model, performs inference, and determines the delivery information based on an inference result.

The information delivery apparatus according to claim 6 is the information delivery apparatus according to claim 1, wherein the delivery information determining unit corrects the 1 st inspection data group and the 2 nd inspection data group in accordance with the 1 st inspection data group acquired by the 1 st device and the 2 nd inspection data group acquired by the 2 nd device, inputs the corrected inspection data groups to the inference model, determines the inference results by the inference models in an integrated manner, and determines the delivery information based on the determination results.

In the information delivery device according to claim 7 of the invention 1, the 1 st inspection data acquisition unit and the 2 nd inspection data acquisition unit determine whether the inspection data set is from the subject person or from a person other than the subject person, and acquire the inspection data set as the 1 st inspection data set or the 2 nd inspection data set when the inspection data set is from the subject person.

The information transmission device according to claim 8 is the information transmission device according to any one of the above-described 1 st to 7 th inventions, wherein the 1 st and 2 nd test data sets are data obtained by outputting a result based on any one of a color sensor, a shape sensor, a hardness sensor, an olfactory sensor (including nematode or animal reaction judgment), a gas component sensor, a color change detection sensor when a specific reagent is added, and shape judgment based on an enlarged observation image, which are used for defecation.

In the information delivery method according to the 9 th aspect of the present invention, the 1 st examination data group in time series of the subject person is acquired by the 1 st device, the 2 nd examination data group in time series of the subject person is acquired by the 2 nd device, the 2 nd device can perform an examination capable of complementing the 1 st examination data group, and the delivery information to be provided to the subject person is determined using the 1 st examination data group and the 2 nd examination data group, the 1 st examination data group and the 2 nd examination data group mutually complementing an examination time or an examination item.

ADVANTAGEOUS EFFECTS OF INVENTION

According to the present invention, it is possible to provide an information delivery apparatus and an information delivery method capable of grasping an accurate health state in consideration of the condition of a subject person and providing customized information such as advice corresponding to the health state.

Drawings

Fig. 1 is a block diagram showing the configuration of an information delivery system according to an embodiment of the present invention.

Fig. 2 is a graph showing a time-series change in examination data of a subject person in the information delivery system according to the embodiment of the present invention.

Fig. 3 is a graph showing a time-series change in test data of a target person in the case where a plurality of devices for measuring test data of a target person are provided in the information delivery system according to the embodiment of the present invention.

Fig. 4 is a flowchart showing an example of an operation of transmitting a check result in the information delivery system according to the embodiment of the present invention

Fig. 5 is a flowchart showing another example of the operation of transmitting the inspection result in the information delivery system according to the embodiment of the present invention.

Fig. 6 is a flowchart showing an operation of performing matching adjustment and inference on history data in the information delivery system according to the embodiment of the present invention.

Fig. 7 is a flowchart showing an operation of inference for creating an inference specification in the information delivery system according to the embodiment of the present invention.

Fig. 8 is a flowchart showing an operation of creating an inference model in the information delivery system according to the embodiment of the present invention.

Fig. 9 is a graph showing a time-series change in inspection data of a subject person in a case where a plurality of devices for measuring the inspection data of the subject person are provided in a modification of the information delivery system according to the embodiment of the present invention.

Fig. 10 is a flowchart showing an operation of inference for creating an inference specification in a modification of the information delivery system according to the embodiment of the present invention.

Detailed Description

Hereinafter, an example in which the present invention is applied to an information delivery system will be described as an embodiment of the present invention. In the present embodiment, the following information delivery system will be described as an example of providing customized information while grasping an accurate health state in consideration of the situation of a subject: it is possible to monitor the examination data related to the health status by the 1 st device or the 2 nd device or the like every day and provide the health-related information based on the information. The information delivery system monitors examination data relating to the health status of a subject person using a plurality of devices on a daily basis. If information is collected only by a single device, there is a restriction that information is collected only by the device, and the collected information is limited. In addition, if the device is a separate device, if an error or the like occurs in the device or in restrictions specific to the device such as the usage environment or installation environment of the device, the error or the like affects the determination. In the present embodiment, the inspection data is collected from a plurality of devices, and the matching adjustment is performed on the inspection data to determine the matching. As a result, it is possible to grasp the accurate health state of the subject person in response to a situation in which accuracy cannot be improved in a single facility.

That is, in the case of acquiring inspection data by a plurality of apparatuses, compared with the case of performing an inspection by forcibly using the same apparatus, there is no burden on the user, data is easily acquired, and data can be provided unconsciously, which is convenient. In addition, since the inspection data can be acquired in various situations, the data amount can be increased. For example, when the pressure is dependent on the environment, such as when the blood pressure is raised only in the workplace, or when the physical condition changes depending on the season or the passage of one day, it is preferable to effectively use data of various devices. That is, the health status of the user can be grasped in more detail by processing the data as the examination data set in order to effectively utilize the data of various devices. Further, it is possible to identify which device the specific person is from based on information acquired from the specific device within the specific time period, by color classification or the like on a common time axis, and if a graph showing the transition of the information value is created, it is possible to grasp the trend of the health state of the person.

That is, even in a normal chart, if data changes of different items are displayed simultaneously in different colors on the same chart, as compared with a chart in which data changes of the same item are displayed only, overall information may be obtained. Then, if the 1 st inspection data group and the 2 nd inspection data group are acquired so as to complement each other with the inspection time or the inspection item and expressed in a comprehensive (comprehensive) comprehensive expression in a recognizable manner, judgment or determination can be made using the expression, and inference can be made by an inference model using the expression. It is also possible to make a judgment or a decision based on the inference result after obtaining the inference result, and associate the judgment or decision result with another control. The judgment and determination may be performed by a rule or pattern matching method. As an inference method using deep learning or the like, it is sufficient to simply represent a graph itself as an image, prepare an inference model using the image as a result of learning as training data, and input the same graph to the inference model to perform inference.

However, in this case, the measurement result may vary depending on facility differences, specification differences, installation environments, and the like of the plurality of facilities. Therefore, measures are taken against the measurement variation, and the advantage of a large data amount is effectively utilized. Further, the user's situation at the time of examination is taken into consideration, and customization information such as a suggestion in which the situation is taken into consideration can be generated and notified to the user.

In the information delivery system according to the present embodiment, since the inspection data set including the time-series inspection data acquired by the different devices is dependent on the devices and the environment, even if the levels of the respective inspection data are different, if the data are regarded as the time change patterns of the same biological information, the data tend to be the same for the same person. Then, by correcting each inspection data group by addition-subtraction or multiplication-division, the variation of the inspection data when the inspection is performed in a plurality of devices is eliminated, and the levels of the respective data in each inspection data group are made substantially equal (for example, see the graph 34 in fig. 3). By performing this correction, a plurality of inspection data sets can be handled as a single inspection data set. Therefore, the data can be increased, and the accurate health state of the subject person can be grasped.

Further, by designing to change the meaning of data or to make data weighting have a difference or the like for each device, the difference of devices can be reduced. If the number of devices to be evaluated is increased, the number of data items to be evaluated in time series can be increased, and the influence of errors in the respective devices can be reduced. Further, if it is known which data (including time information) is originated from which device, even if reliability of a certain device is reduced at a specific time due to some cause, it can be designed to perform determination by using only data before the specific time and using the data thus used. For example, when abnormal data occurs due to a failure of a specific device or the like, there is a possibility that a problem that error information is provided to a user using the device, but the problem can be prevented by referring to the result of another device. In addition, when it is possible to determine which device the user uses frequently, for example, a usage pattern can be adopted in which the determination is exclusively performed only by the device and the inspection result of using another device is reflected as necessary. In this case, by performing this "reflection", it is possible to prevent the problems inherent to the apparatus.

In the information delivery system according to the present embodiment, regarding variations in inspection data when a plurality of devices are inspected, each inspection data set is corrected by addition and subtraction or multiplication and division so that the levels of the respective data in each inspection data set substantially match, and the corrected inspection data set is input to the inference model, whereby the reliability of inference at that time is calculated (see, for example, S73 and S77 in fig. 10). When the correction amount in the correction calculation is changed little by little, the reliability also changes little by little. The information delivery system uses the inference result when the reliability is highest.

In the information delivery system according to the present embodiment, the reliability is calculated by changing the correction amount in the correction calculation little by little, and the inference result when the reliability is the highest is used as the inference result of each inspection data group. The information delivery system comprehensively makes a judgment based on each inference result, and takes the judgment result as an inference result (for example, refer to fig. 9).

The subject person in the present embodiment is a person who may become a patient through a review. The subject is also a person who is reassuring health, does not worry about diseases, and can enjoy daily life according to the result of the review. In addition, the health can be similarly improved by simple life improvement, treatment, or the like.

The information delivery system according to the present embodiment is configured by, for example, a server, but may be configured by a portable information device such as a personal computer or a smartphone that can exchange information with the server.

In the present embodiment, when determining delivery information to be delivered to a subject, a search and/or inference are performed based on the result of an examination by the subject, and equipment necessary for performing a further examination or the like based on the result of the examination, and facilities having the equipment. For this search/inference, a Database (DB) storing facilities with equipment is preferably set in advance. In addition, when providing the delivery information, information such as an access method including a facility name, a telephone call, an e-mail, a map, and the like, a visit time or an idle time, and a charge estimate may be included. Note that the facility is not limited to one, and may be plural.

Next, a configuration of an information delivery system according to an embodiment of the present invention will be described with reference to fig. 1. The information delivery system is composed of a control unit 1, a 1 st device 2a, a 2 nd device 2b, a 3 rd device 3, a terminal 4, a learning unit 5, a learning delegation unit 6, an inference engine 7, a Database (DB) unit 8, and a correlation check mechanism (including a medical institution or the like) 9. In each of these units, the control unit 1 is disposed in a server, and the 1 st device 2a, the 2 nd device 2b, the 3 rd device 3, the terminal 4, the learning unit 5, the learning delegation unit 6, the inference engine 7, the DB unit 8, and the association check mechanism 9 can be connected to the server via a network such as the internet. However, the present embodiment is not limited to this configuration, and for example, one or more of the control unit 1, the 1 st device 2a, the 2 nd device 2b, the 3 rd device 3, the learning unit 5, the learning delegation unit 6, the inference engine 7, and the DB unit 8 may be disposed in a server, and the other may be disposed in another server, an electronic device such as a personal computer, or the like. The association check means 9 may also have a server function.

The control Unit 1 is a controller (Processor) for controlling the information delivery system according to the present embodiment, and assumes an IT device including a CPU (Central processing Unit), a memory, an HDD (Hard disk Drive), and the like, such as a server, and providing files, data, and the like to other terminals via a network. However, the control unit 1 is not limited to this configuration, and may be configured by a device such as a personal computer when constructed as a small-scale system. The control unit 1 has various interface circuits, can cooperate with other devices, and can perform various kinds of arithmetic control by a program.

The control unit 1 receives information from each apparatus performing cooperation, collates the information, generates necessary information, and provides the information to the user. The control unit 1 also has a function of outputting a request to each device that cooperates with each other and operating each device. In the present embodiment, it is expected that the degree of freedom of the system is high and the convenience of use is high, and the device 1 such as the 1 st device 2a or the terminal 4 or the like of the subject person (also referred to as a user) can be connected to the control unit 1 by wireless communication or wired communication. As communication for this purpose, a wireless LAN or a cellular phone communication network is also conceivable, and short-range wireless such as bluetooth (registered trademark) or infrared communication is used in combination depending on the situation. Since the description of the communication unit including the communication circuit, the antenna, the connection terminal, and the like is complicated, it is omitted in fig. 1, but a communication unit including the communication circuit and the like is provided in an arrow portion showing communication in the drawing.

The control unit 1 includes a communication control unit 1a, an ID determination unit 1b, an information providing unit 1c, an inference model specification determination unit 1d, an inference delegation unit 1e, and a search unit 1 f. These respective sections may be realized by software based on a CPU, a program, or the like in the control section 1, may be realized by a hardware circuit, or may be realized by cooperation of software and a hardware circuit. In fig. 1, the direction of a signal for realizing each function by the cooperation of each part in the control unit 1 is omitted, but the description is given in the flowchart. For example, in the step of S1 in fig. 5, the ID determination unit 1b collects information from the 1 st device 2a, the 2 nd device 2b, and the like for each identical user.

The communication control unit 1a has a communication circuit and the like, and performs transmission and reception of data and the like with communication units provided in the 1 st device 2a, the 2 nd device 2b, the 3 rd device 3, the terminal 4, the learning unit 5, the learning delegation unit 6, the inference engine 7, the Database (DB) unit 8, and the association check means 9. Each device/unit such as the 1 st device 2a, the 2 nd device 2b, the 3 rd device 3, and the terminal 4 also has a communication unit, but the illustration thereof is omitted because it becomes complicated in fig. 1.

The ID determination unit 1b collects information from the 1 st device 2a and the like for each identical user. An ID is assigned to each person in order to identify the person who has acquired information by the 1 st device 2a, the 2 nd device 2b, the 3 rd device 3, and the association check means 9. In the present embodiment, since data of each user is processed, the ID determination unit 1b manages which user receives information and which user provides guidance. The specific user is determined by providing the 1 st device 2a, the 2 nd device 2b, and the 3 rd device 3 with a biometric authentication function, inputting an ID by the user through the terminal 4, transmitting an ID by the user through the communication unit in the 1 st device 2a and the 2 nd device 2b, or reading a unique code by the terminal 4. In addition, in order to protect personal information, management is enhanced by encrypting necessary portions, but since these are general techniques, detailed description thereof is omitted.

The ID of each device may include information relating to the model name of the device or specific information indicating which individual the device belongs to. The function, performance, and the like of the mounted sensor may be known by the model name, the installation location, the use environment, and the like may be known by the individual information, and the information may be searched via a network or the like. If the model name is known, information of similar devices can be determined, and latitude and longitude, indoor and outdoor, season, climate, temperature characteristics, and the like can be determined according to the installation place and the environment, and the output information of the device can be corrected in consideration of the determination result.

The ID determination unit 1b functions as a 1 st inspection data acquisition unit that acquires a 1 st inspection data set of the subject person as a time series by the 1 st device (see, for example, S101 in fig. 4 and S1 in fig. 5). The ID determination unit 1b also functions as a 2 nd inspection data acquisition unit that acquires a 2 nd inspection data set of the subject in time series by a 2 nd device (see, for example, S105 in fig. 4 and S1 in fig. 5), and the 2 nd device can perform an inspection for complementing the 1 st inspection data set. The ID determination unit 1b functions as a 2 nd inspection data acquisition unit that acquires a 2 nd inspection data set as a time series of the subject person by a 2 nd apparatus capable of performing the same inspection as the 1 st apparatus. The 1 st inspection data set and the 2 nd inspection data set complement each other in inspection time or inspection items.

The 1 st inspection data acquisition unit and the 2 nd inspection data acquisition unit determine whether the inspection data set is from a subject person or from a person other than the subject person, and acquire the inspection data set as the 1 st inspection data set or the 2 nd inspection data set when the inspection data set is from the subject person. The acquired 1 st and 2 nd inspection data sets are recorded in a recording unit. The test data set other than the subject person may be recorded in advance so that the training data can be formed in association with the health information of the person other than the subject person.

The information providing unit 1c has a function of acquiring information of the user (may refer to a result acquired by another device) in order to provide accurate information to the user. The information providing unit 1c acquires the inspection data of the user (specified by the ID) acquired from the 1 st device 2a or the like or the related inspection means 9. Further, the information providing unit 1c determines the health status of the user using the acquired examination data, various information acquired from the association examination means 9, information about the held device stored in the DB unit 8, profile information of the user, and the like. As the health state, a current disease or a disease in which symptoms may appear in the future is included, and when the health state is judged, information associated with the health state is provided to the user. In addition, when a disease or the like of the user is determined, information on facilities to be subjected to examination or treatment is provided to the user as necessary.

Further, if the control unit 1 can inquire of the correlation check means 9 about the current hospital visit status, information such as prescription drugs, past health diagnosis results, and the like based on the ID of the user and the like in order to confirm the health status of the user in the specific status, the judgment of the correlation with the device data becomes easy. The user operating the terminal 4 can deal with the problem of safety by allowing the cooperation or performing an operation for offering cooperation by a doctor operating the (IT device of the) related examination mechanism 9.

That is, the information providing unit 1c provides the user with information relating to health, for example, information indicating when the user accesses a facility to receive an examination or a treatment, and information for recommending a facility suitable for receiving an examination or a treatment. The information providing unit 1c acquires the inspection data transmitted from the 1 st device 2a or the like or the related inspection means 9. As described later, this data is inspection data (time series information) to which time information is added, and is stored in a data structure that can be formed into a graph as shown in fig. 2 and 3. In the present embodiment, it is assumed that the control unit 1 provides information to the user using information from the 1 st device 2a or other devices in the association check means 9, but a modification in which information is collected by a server having the association check means 9 in the same manner may be used.

In order to provide such information, the information providing unit 1c collects the inspection data from the 1 st device 2a, the 2 nd device 2b, and the like, and records the data in the DB unit 8. The frequency of information acquisition and the number of data may be different depending on the 1 st device 2a, the 2 nd device 2b, and the like. That is, the increase and decrease of the specific health-related value obtained in each facility are arranged in time series, and the value measured by changing the facility can be arranged for each facility.

For example, even if the blood pressure is the same, examination data measured by a wearable simple device and examination data measured by a dedicated device at an event venue are recorded separately in advance. If it is assumed that the wearable device is worn almost all the time, the examination data increased or decreased according to the state is obtained in the morning, evening, before meals, after sleeping, before getting up, and after getting out of bed. Further, although the inspection data measured by the dedicated apparatus is single, data with higher accuracy or in a form of accompanying information (such as an inspector, subjective symptoms heard by an inspector, and the like) is obtained. In addition, when measurement is performed by a dedicated device, the measurement data is accurately acquired because clothes are especially removed or no meal or dedicated meal is eaten. In such a situation, the difference in the position or item of the human body to be measured is small, and the error of the equipment is strictly controlled, and it is appropriate to compare the absolute value with another person.

The information providing unit 1c may acquire a lifestyle, a eating habit, a sleeping time, a dining time, and the like of the user at a residence or a work place on the internet, and may generate information such as facilities provided to the user in consideration of the acquired information. The acquisition of such information can be supplemented by general or widely known techniques. Further, the information providing unit 1c may customize information such as facilities generated by acquiring the information. The information on the relevant facility is acquired from the correlation inspection means 9 as medical institution information.

The information providing unit 1c uses information such as the held device stored in the DB unit 8 in addition to information collected from the association check means 9 such as the 1 st device 2a when providing information such as recommended facilities. Of course, the information recorded in the DB8 may be recorded in a different recording unit other than the DB unit 8. In this case, a plurality of DB units in fig. 1 are not shown because they are complicated. The information providing unit 1c collects various information when providing information. That is, the information providing unit 1c functions as an acquiring unit that acquires the examination data of the user, the material information of the user, and the held device information for each examination/medical institution.

The information providing unit 1c functions as a delivery information determining unit that determines delivery information to be provided to the target person using the 1 st and 2 nd inspection data groups (see, for example, S107 in fig. 4, S9 in fig. 5, S79 in fig. 10, and the like). The delivery information determining unit determines the delivery information according to an inference model learned according to the change pattern of the inspection data set acquired by the plurality of devices (for example, see S107 in fig. 4, S9 in fig. 5, S79 in fig. 10, and the like).

The delivery information determining unit corrects the 1 st inspection data group and the 2 nd inspection data group according to the 1 st inspection data group acquired by the 1 st device and the 2 nd inspection data group acquired by the 2 nd device, inputs the corrected inspection data groups, calculates the reliability of the inference result at that time, and determines the delivery information according to the reliability (see, for example, S107 in fig. 3 and 4, S9 in fig. 5, 9, and 10). When performing correction for each inspection data group, the transmission information determination unit performs four arithmetic operations on numerical values common to the respective data included in the inspection data group.

As the four arithmetic operations, for example, in the case of devices that acquire the same biometric information, the reliability of the inference may be calculated while performing the correction of different devices, and the result of high reliability may be used as the inference result. When four arithmetic operations are performed, the constants of the specific four arithmetic operations are changed little by little for time-series data of different devices, and the four arithmetic operations are performed. This process increases the reliability in a situation where the error is corrected, and therefore, accurate inference can be performed. With such a design, if the devices acquire the same biological information, it is possible to provide information with reliability regardless of errors or noises that may be influenced by the sensitivity or the use environment. Further, the same biological information is not necessarily required, and even if different data such as pulse, heartbeat, and respiratory rate are used, if similar changes are made including the magnitude relationship of the data, it is possible to comprehensively perform accurate inference by correction and reliability determination.

The delivery information determination unit corrects the 1 st inspection data group and the 2 nd inspection data group in accordance with the 1 st inspection data group acquired by the 1 st device and the 2 nd inspection data group acquired by the 2 nd device, merges the corrected inspection data groups into 1 inspection data group, inputs the merged inspection data group to the inference model, performs inference, and determines delivery information based on the inference result (for example, refer to the graph 34 in fig. 3, S107 in fig. 4, and the like). The delivery information determining unit corrects the 1 st inspection data group acquired by the 1 st device and the 2 nd inspection data group acquired by the 2 nd device in accordance with the 1 st inspection data group and the 2 nd inspection data group, inputs the corrected inspection data groups to the inference model, comprehensively determines the inference results based on the inference models, and determines the delivery information based on the determination results (see, for example, fig. 9 and 10).

As described above, the information providing unit 1c acquires the inspection data in the time-series pattern that becomes the specific period of the user. The acquired time-series pattern is not constituted by data obtained by only one measurement, but is constituted by each of inspection data obtained by measurements at a plurality of different times, and changes in the pattern of the inspection data are also used as information. By using a time-series pattern composed of a plurality of pieces of inspection data, it is less susceptible to errors due to changes in the measurement environment or conditions. Further, the state of health from the end time of the specific period to the future time (the time when the specific period is extended) is inferred, and prediction for the future can be performed.

In addition, if the time information of the user coming to the examination/medical institution is given as the label information with respect to the acquired time-series pattern, the training data can be obtained. If there is an inference unit having an inference model generated by learning using the training data, it is possible to infer what occurs at a time (when the specific period is extended) after the specific period (period for acquiring the time-series change pattern). Further, if the name of a disease or the like of the user is known, it is possible to generate training data to which the information is given as labeling information. By learning using the training data, an inference model for inferring health information such as a disease can be generated. Here, in generating the inference to be used, the specification of the specific input/output information is defined and learned.

Therefore, in the present embodiment, the time-series change pattern of the user's examination data is input to the inference unit, the inference unit performs inference, and the transmission information determination unit is provided which determines the transmission information at a time after the specific period based on the inference result. Therefore, it is possible to provide a system, an apparatus, a method, a program, and the like capable of transmitting prediction information of a time after the time of acquisition of the time-series pattern.

If there is a difference in mechanical performance or the like in the respective inspection apparatuses, the reliability of the inspection data of the user is degraded. For example, when the same user acquires examination data (biological information) by using a plurality of devices at the same time, the same examination data may not be acquired. Therefore, if a plurality of inspection apparatuses (inspection apparatuses of a specific specification) capable of inspecting the same inspection item are used to repeatedly perform inspections at different dates and times to acquire change pattern information of a large amount of inspection data, it is possible to handle the change pattern information as large data. In this case, the specific period need not be a fixed period, but may be a different time width (specific period) depending on the situation. The "time width" as used herein is not a time (inspection interval, measurement interval) between the measurement time and the measurement time, but refers to a time interval from the first measurement to the last measurement when a series of inspection data is acquired. The "time width" may be rewritten to a specific time width including change pattern information of the inspection data, in which a large amount of time-series data is included in the time width.

In the present embodiment, the information providing unit 1c inputs a change pattern of the inspection data to the inference engine 7 in which the inference model generated by the learning unit 5 is set, obtains an inference result related to the advice, and provides the inference result to the user corresponding to the input inspection data. This service may use personal information, and in order to receive a provision of advice or the like, it may be necessary to match the personal information. In this sense, the profile information of the user is sometimes important. In addition, when the user is an infant or an elderly person, the advice may be delivered to a person who cares for the user, a caregiver, or the like. In this case, valid information such as advice is also delivered in accordance with information managed in the profile information of the user.

The inference model specification determining unit 1d determines the specification of the inference model to be generated when the inference requesting unit 1e requests the learning unit 5 to generate the inference model by the learning requesting unit 6. The control unit 1 acquires the biometric information of the user from the 1 st device 2a and the like, and stores the biometric information. The control unit 1 requests the learning unit 5 to generate various inference models by the learning requesting unit 6 using the stored biometric information as training data. The inference model specification determining unit 1d determines what specification of inference model is requested when generating the inference model. For example, as shown in fig. 2(a) described later, when the time-series biological information is stored, the inference model specification determining unit 1d determines the specification of an inference model for inferring what kind of examination data (value) the user receives treatment through the medical facility several days later. The inference model specification determining unit 1d determines, based on the time-series biological information, a specification for generating an inference model that infers what kind of disease is currently suffered from, what possibility of suffering from what kind of disease in the future (when), and a facility recommended for receiving a required examination or treatment when a disease is possibly suffered from.

The inference delegation unit 1e delegates the generation of the inference model of the specification determined by the inference model specification determining unit 1d to the learning unit 5 through the learning delegation unit 6. That is, when a predetermined number of pieces of biological information acquired by the 1 st device 2a and the like are stored, the inference requesting unit 1e requests the learning unit 5 to generate the inference model by the learning requesting unit 6, and receives the generated inference model by the learning requesting unit (or directly from the learning unit 5). The received inference model is sent to the inference engine 7. Further, the control unit 1 preferably prepares a plurality of inference models, and appropriately selects an inference model based on information to be provided to the user.

The search unit 1f searches an examination facility or a medical facility having a facility necessary for examination or treatment in the database stored in the DB unit 8 when it is found that a disease currently exists, there is a possibility of what disease exists in the future (when), and examination or treatment is necessary, based on the biological information of the user acquired by the 1 st apparatus 2a, the 2 nd apparatus 2b, and the 3 rd apparatus. These pieces of information can be obtained by inference using the inference engine 7, but may sometimes be matched with stored data. Since there are cases as well, in the present embodiment, the search can be performed by the search unit 1 f.

The 1 st device 2a and the 2 nd device 2b are devices for acquiring examination data of health-related information of the user, for example, vital sign information, sample information, and the like. The 1 st device 2a and the 2 nd device 2b are inspection devices of specific specifications, and are devices capable of inspecting the same kind (same) of health-related information. When the inspection data sets acquired by the 1 st device 2a and the 2 nd device 2b have different inspection times, it is sufficient if inspection is possible to complement the two types of data. The 1 st device 2a and the 2 nd device 2b may not check the same items, and for example, even when the heart rate is measured while measuring the blood pressure, the two types of data may be complemented with each other. In fig. 1, only 2 devices, i.e., the 1 st device 2a and the 2 nd device 2b, are shown as devices for acquiring the user's examination data, but the number is not limited to 2, and may be 3 or more. As described later, the 3 rd device 3 is assumed in the present embodiment as a device for acquiring examination data of a person other than the user.

The health-related information acquired by the 1 st device 2a and the like includes various information, for example, vital sign information such as body temperature, blood pressure, and heartbeat of the user. The health-related information includes various kinds of sample information such as urine, feces, and the like, sputum, blood, and the like of the user. In the case of stool, the 1 st device 2a and the 2 nd device 2b acquire the color, shape, amount, date and time information thereof. The 1 st device 2a and the 2 nd device 2b may acquire information in accordance with an instruction from the control unit 1, may acquire information in accordance with an operation by a user, and may automatically acquire information. The 1 st device 2a and the like may collect and use a Personal Life Record (PLR) obtained by adding various activity data of daily Life such as activities in daily Life, workplace/school, meals, and sports activities to the information "Personal Health Record (PHR)", which is medical Health information, and the acquired information may be transmitted to the control unit 1 through a communication unit (not shown) in the 1 st device 2a and the like.

The inspection data of the subject person detected by the 1 st equipment 2a and the like is obtained by acquiring the inspection data in time series using an inspection equipment of a specific specification and extracting change pattern information of the inspection data in a specific time width. That is, as the 1 st equipment 2a and the 2 nd equipment 2b, inspection equipment of a specific specification (inspection equipment of the same type) is used, and the 1 st equipment 2a and the like measure inspection items of the same subject person at different times, thereby acquiring data in time series. The measurement value is plotted on a graph according to the examination time using the time-series data, thereby obtaining a change pattern. The inspection data set can be obtained by extracting the variation pattern in a specific time width. The examination data is data obtained by outputting a result of any one of a color sensor, a shape sensor, a hardness sensor, an olfactory sensor (including reaction judgment of nematodes or animals), a gas component sensor, a color change detection sensor when a specific reagent is added, and shape judgment based on an enlarged observation image, which are used for defecation.

When the 1 st device 2a and the 2 nd device 2b obtain information related to the specific user, the information providing unit 1c of the control unit 1 presents information related to recommended facilities to the information terminal 4 of the specific user. The description is given assuming that the presentation assists the user's action, but various modifications can be considered.

The information determination performed in the 1 st device 2a and the like may be changed to what degree according to the relationship with the control unit 1. For example, only the result sensed in the 1 st device 2a or the like may be transmitted to the control unit 1 without determination. However, in this case, it is necessary to add information indicating what kind of data of what kind of person is to the sensing signal and transmit the signal. The attached information preferably corresponds to which person or which sensing result, but may correspond to information of other terminals by being added to information of other terminals.

The 3 rd device 3 is a device that acquires data of a person different from the user using the 1 st device 2a or the 2 nd device 2 b. In fig. 1, only 13 rd device 3 is described, but a plurality of devices may be provided, and an unspecified number of devices are collectively shown in fig. 1. This makes it possible to record and manage what kind of person is what kind of disease and what kind of health value is as big data.

The 3 rd device 3 composed of the unspecified number may also take different values with different performances. The more such devices that are added to the system as health management devices, the more data that can be utilized as health monitoring devices. In an extreme example, if the result of each person taking a self-timer with a smartphone every day is combined with other health values of the person to form big data, data such as facial color deterioration from how long ago the person had a disease can be effectively used. If this data is used effectively, early health management, continence, treatment can be suggested to others in the presence of similar facial color changes.

When wearable terminals are used as the 1 st device 2a, the 2 nd device 2b, and the 3 rd device 3, vital sign information such as body temperature, heartbeat, blood pressure, brain waves, line of sight, breathing, and exhalation can be obtained by bringing the wearable terminals into close contact with the skin or the vicinity of the body through the wearing part of the wearable terminals. In addition, as a weighing scale, a sphygmomanometer, and a measuring instrument for measuring arteriosclerosis indicating the hardness of an artery wall, dedicated precision equipment is disposed in a health facility, a public bathing place, a pharmacy, a shopping center, and the like, and a professional measurer may be disposed together. In such facilities, users often use measurement devices easily at idle time or the like, and manage their physical conditions based on the measurement results at that time. These measuring apparatuses may be referred to as the 1 st apparatus 2a, the 2 nd apparatus 2b, and the 3 rd apparatus 3.

The 1 st device 2a, the 2 nd device 2b, and the 3 rd device 3 may request to record a questionnaire before or after the user uses a dedicated terminal or the like. In such a case, the profile information or other information of the user can be specified based on the description of the questionnaire. Such information collection is not limited to the 1 st device 2a and the like, and may be performed by the control unit 1. This information can be used when determining whether or not specific information is acquired in step S3 of fig. 5, which will be described later. If the information of when the doctor is seen can be heard, it can be used as the time Tc information in fig. 2(a) and 2(b) described later.

The 1 st device 2a, the 2 nd device 2b, the 3 rd device 3 may also be a thermometer, a sphygmomanometer or the like that has suffered from a specific illness and is used under the direction of a doctor. In addition, in the case where colors of faces, nails, and the like, facial expressions, images of affected parts, sounds when the throat is uncomfortable, and the like, which are photographed by a camera provided in the smartphone, are picked up by a microphone, the portable terminal (smartphone) can be directly used as the 1 st device 2a, the 2 nd device 2b, and the 3 rd device 3.

Recently, a simple health management device and a health information acquisition device have been developed, and these devices are sometimes mounted on a wearable device, and many of these devices are handled as peripheral devices of a smartphone, not as standalone devices, and therefore they can be assumed to be portable terminals. In addition, even if the wearable device is not used, a simple measurement device may be installed in a place where people gather to provide a health information service. Such devices may also be used as the 1 st device 2a, the 2 nd device 2b, the 3 rd device 3.

The related examination means 9 is a facility where the user receives an examination, and includes, for example, an examination facility or a medical facility. The related examination means 9 may be of a mobile type, for example, a type in which a general medical facility or an examination facility is mounted on an automobile, a train, a ship, a helicopter, an unmanned aerial vehicle, or the like and goes to the patient. The control unit 1 can acquire which medical facility is going to and what examination result is obtained from a server or the like of the system that operates the relevant examination facility 9. Of course, the server of the association check mechanism 9 may be the same as the control unit 1, and may share some functions.

As described above, the terminal 4 is a portable information terminal and is a device for receiving information that can be confirmed by the user or a person related thereto. The information includes health information and facilities recommended according to the health status. The terminal 4 may be, for example, a smartphone or a tablet PC, and in this case, a built-in camera or a microphone may be used as the information acquisition unit. Further, another home appliance which is a wearable terminal capable of cooperation may be used as the terminal 4, and information may be acquired by the wearable terminal or the like. Therefore, the 1 st device 2a or the 2 nd device 2b may also be the same as the terminal 4 and may also be dedicated devices, respectively. The terminal 4 cooperating with the wearable terminal may acquire information and manage the information. Depending on the situation, the 1 st device 2a, the 2 nd device 2b, the 3 rd device 3, or the terminal 4 may have the function of the control unit 1, or may be configured to perform detection, control, and information provision in a shared manner.

The Database (DB) unit 8 has an electrically rewritable nonvolatile memory. The DB unit 8 has a data history list of different IDs, and this list records the relationship between the acquired data and the examination day. As described above, since the ID determination unit 1b receives the inspection data from the 1 st device 2a or the like or the related inspection means 9 or the like, the DB unit 8 records the inspection data for each ID. At this time, the inspection date, the inspection equipment (1 st equipment, 2 nd equipment, 3 rd equipment, associated inspection means, or the like), the inspection location, the inspection item, and the like are also recorded. In addition, how and for what the inspection is performed, etc. are also recorded. The DB unit 8 may sort the acquired data into 5W1H, that is, WHO (WHO), WHERE (WHERE), WHEN (date and time), WHAT check, WHAT, HOW, and record the sorted data.

The DB unit 8 may include a storage facility recording unit that records a list of storage devices of different facilities, and a reception history recording unit that records IDs and the reception of the user for each facility. The storage facility recording unit records a list of devices stored in facilities such as hospitals, clinics, and examination facilities. The information providing unit 1c can present information on the facility having the most suitable device for inspection to the user by searching the stored facility recording unit. The information may be coordinated with the information of the relevant inspection means 9 so as to update the information according to the situation of replacement of the device such as a medical facility. Further, the arrival history recording unit records, for each facility, the arrival information of who (determined by the ID) is at what time.

The DB unit 8 is a part of an information transmission system that cooperates with a medical facility, and the DB unit 8 may be accessible to the relevant inspection means 9 through the control unit 1. In this case, when the DB unit 8 receives a search command from the control unit 1, the DB unit 8 searches not only the data recorded in the DB unit 8 but also the data in the association check means 9 and outputs the search result. The DB unit 8 functions as a storage unit for storing information on the user profile and information on the equipment held in each examination/medical institution. The storage unit is not limited to the DB unit 8, and all or part of the functions may be disposed in the control unit 1 or the like.

When receiving a request for generating an inference model from the inference requesting unit 1e in the control unit 1, the learning requesting unit 6 transmits specifications of the inference model and the like to the learning unit 5, and requests generation of the inference model in accordance with the specifications. The learning delegation unit 6 includes a data classification recording unit 6a, a specification setting unit 6d, a communication unit 6e, and a control unit 6 f.

The control Unit 6f is a controller (Processor) that controls the inside of the learning delegation Unit 6, and assumes an IT device, such as a server, which is configured by a CPU (Central processing Unit), a memory, an HDD (Hard disk Drive), and the like, which provides files, data, and the like to other terminals via a network. However, the control unit 6f is not limited to this configuration, and may be configured by a device such as a personal computer when constructed as a small-scale system. The control unit 6f has various interface circuits, can cooperate with other devices, and can perform various kinds of arithmetic control by a program.

The data classification unit 6a has an object type a image group 6b, in which training data 6c is recorded. The object type a image group 6B is an image group used when the inference model is generated in the learning unit 5, and includes a large number of image groups of type a and type B …. Training data 6c is generated based on the image group. That is, as shown in fig. 2 and 3, when drawing the examination data for each examination day, a graph can be drawn and the graph can be handled as an image. The data record classification unit 6a records training data 6c based on the data history list recorded in the DB unit 8.

The specification setting unit 6d sets what inference model is to be generated, based on the specification of the inference model determined by the inference model specification determining unit 1 d. In order to satisfy this specification, training data is generated from data recorded in the history list in the DB unit 8.

The communication unit 6e has a communication circuit for communicating with the control unit 1 and the learning unit 5. The communication unit 6e receives a request for generating an inference model from the control unit 1, and the request learning unit 5 generates the inference model.

The learning unit 5 includes an input/output modeling unit 5a, and generates an inference model by machine learning or the like in accordance with the specification from the learning delegation unit 6. The input/output modeling unit 5a includes a specification matching unit 5 b. The specification comparison unit 5b determines whether or not the specification received from the learning delegation unit 6 matches the inference model generated by the input/output modeling unit 5 a. That is, the specification comparing unit 5b defines not only the input/output relationship but also a learning method and the like so as to learn in accordance with the "required specification" such as the time, energy, and circuit configuration required for the inference of the inference model.

The inference model is generated by learning the relationship between acquired information such as acquired biological information and biopsy information and a disease, specifically, the relationship between the acquired information and a department of medical science. The input/output modeling unit 5a has an input layer, a plurality of intermediate layers, and an output layer, as in the case of the inference engine 7, and obtains the coupling strength of neurons in the intermediate layers by learning to generate an inference model.

When generating such an inference model, the learning delegation unit 6 extracts a change pattern of the examination data acquired from the examinee using the examination device for a specific time width, inputs the extracted change pattern to the inference engine 7, and generates training data in which a health advice to be output at a time after the examination time of the examinee is generated as annotation information. Then, the learning unit 5 performs learning by using the training data to generate an inference model. In addition, although the present embodiment has been described with respect to the time width traced from the time point at which the examination result is obtained, there are cases where the data is improved by treatment or the like after the examination result and cases where the treatment is not favorable, and therefore, it is also possible to learn the difference and output a suggestion for prognosis (after illness).

Further, if the learning unit 5 performs learning using the examination data string after examination, hospital visit, and medicine administration, it is also possible to generate an inference model that can make future prediction suggestions such as lifestyle improvement, treatment, and effect of medicine administration. In this case, the subsequent time-series data is used starting from the time point of examination, hospital visit, and medication. When a recommendation is made for examination, hospital visit, medicine taking, and the like, the time-series data up to that point is used.

Here, deep learning will be described as an example of learning performed by the learning unit 5. "Deep learning (Deep learning)" is learning in which the process of "machine learning" using a neural network is structured in multiple layers. A representative example is a "forward propagation type neural network" that transmits information from front to back to make a decision. In the simplest configuration, the forward propagation type neural network may have 3 layers, i.e., an input layer including N1 neurons, an intermediate layer including N2 neurons given as parameters, and an output layer including N3 neurons corresponding to the number of classes to be discriminated. The input layer and the intermediate layer, and the intermediate layer and the output layer are connected by coupling weights, and by applying bias values to the intermediate layer and the output layer, a logic gate can be easily formed.

The neural network may have 3 layers if it is simply discriminated, but by providing a large number of intermediate layers, it is possible to learn a combination of a plurality of feature amounts in the machine learning process. In recent years, 9 to 152 intermediate layers have been practically used from the viewpoint of time required for learning, determination accuracy, and energy consumption. Further, a "convolution type neural network" may be used which performs a process called "convolution" for compressing the feature amount of the image, and operates with a minimum process and has strong pattern recognition. Further, "recurrent neural networks" (fully coupled recurrent neural networks) that process more complex information and analyze information with changing meanings according to order or sequence to flow information in both directions may be used.

In order to realize these techniques, an existing general-purpose arithmetic processing circuit such as a CPU or an FPGA (Field Programmable Gate Array) may be used. However, the Processing of the neural network is not limited to this, and since the Processing is mostly matrix multiplication, a processor called GPU (Graphic Processing Unit) or Tensorial Processing Unit (TPU) dedicated to matrix calculation may be used. In recent years, a "neural Network Processing Unit (NPU)" designed to be able to integrate and assemble such Artificial Intelligence (AI) -dedicated hardware with other circuits such as a CPU may be a part of a processing circuit.

Further, as a method of machine learning, for example, a support vector machine and a support vector regression method are also available. Learning here is a method of calculating the weight, filter coefficient, and offset of the identifier, and in addition, there is a method of using a logistic regression process. In the case of having the machine make certain decisions, humans need to teach the machine the method of decision. In the present embodiment, a method of deriving the image determination by machine learning is adopted, but in addition to this, a rule-based method that adapts to a rule obtained by a human being through a rule of thumb/heuristic (hearistics) may be used.

The inference engine 7 has the same input/output layer and neural network as the input/output modeling unit 5a of the learning unit 5. The inference engine 7 performs inference using the inference model generated by the learning unit 5. For example, the inference engine 7 inputs the biological information measured by the 1 st device 2a or the like as a time series, and finds an examination institution/medical institution suitable for examination, treatment, or the like of the health status of the user by inference, for example. Further, it is also possible to infer when a medical treatment is received in a medical institution based on time-series biological information.

In this way, the control unit 1 may provide information on recommended facilities by the inference engine 7 in addition to the search of the DB unit 8 by the search unit 1 f. The inference engine 7 performs inference of information on recommended facilities using the inference model generated by the learning unit 5. The inference model is generated by learning the relationship between acquired information such as acquired biological information and biopsy information and a disease, specifically, the relationship between the acquired information and a department of medical science. In this way, the control unit 1 can output guidance information to be presented by the inference engine 7.

When the control unit 1 guides medical facilities or the like in one determination based on the acquired information obtained at one time by search or inference, the medical information may be uselessly taken into life to prevent healthy and reassuring life. Therefore, the accuracy can be improved by using a history (time-series information) of the information acquired a plurality of times.

Fig. 2(a) shows a graph using history data (time-series data) related to the health of an individual recorded in the recording unit 8. The recording unit 8 records, for example, data acquired by the specific device a or data of devices having various inspection functions, the specific data a and at which facility the person has been diagnosed. The control unit 1 manages the recording by the recording unit 8.

The horizontal axis of the graph shown in fig. 2(a) represents time, and the vertical axis represents health-related data. The data transferred on the graph can be handled as image data in which information is arranged in two dimensions. Accordingly, it is possible to perform inference by the same method as the image search, such as finding a specific image from the image. That is, the graph may be an input and the output may be a health-related suggestion. As a suggestion, a measurement method, a name of a specific disease currently suffered, a name of a specific disease possibly suffered in the future, guidance of a treatment/examination facility of a specific disease, and the like are assumed. The graph showing the history data shown in fig. 2 will be described in detail later.

If the recording unit 8 records what facility the person whose health relationship information has changed has already gone or is going, and the control unit 1 manages the records collectively, it is possible to collect these data and use them as training data, and cause the learning unit 5 to create an inference model.

The inference engine 7 has an inference model obtained by specifying the specification of the inference model by the inference model specification determining section 1d of the control section 1 and learning according to the specification. The inference engine 7 may have a plurality of inference models because the device data may be different when a new device is present. A plurality of inference models may be prepared in advance, and the inference models may be selectively determined according to the examination data of the user. Further, since new learning is required every time a new device appears, it is assumed that the inference model is improved or newly created by the learning unit 5 under the designation of the control unit 1 in many cases. However, in case the 1 st device 2a etc. is dedicated and dedicated only for inference of a specific lesion, it may also be a separate dedicated inference model.

Similarly to the input/output unit 5a of the learning unit 5, the inference engine 7 also includes a memory or the like in a circuit block centered on an AI chip such as a CPU, a GPU, or a DSP, and configures a neural network. In the present embodiment, it is assumed that the following may exist: the inference engine 7 and the learning unit 5 are connected to a network or the like operated in cooperation with a hospital or the like, and the control unit 1 can be used in cooperation with these. In this case, too, there is a possibility that information of learning or reasoning can be exchanged via the association check mechanism 9.

When it is determined that the inference requesting unit 1e of the control unit 1 can sufficiently acquire information of the specific user from the 1 st device 2a or the like, it requests the inference engine 7 to perform inference. The inference engine 7 can input an information group indicating the passage of time series of similar data, perform inference based on the information group, and output medical institution information (hospital arrival information, examination information, etc.) suitable for a specific user. In the case where a data set similar to a person who frequently goes to a hospital and has a chronic disease is input to the inference engine 7, it is preferable to guide the display of facilities capable of performing the same treatment. The inference engine 7 is able to evaluate the medical institution holding the device that should examine a specific user.

In addition, since the possibility of incorporating a health monitoring function in a device such as a home appliance is increasing in the future, even if a user does not go to a place where a special device is installed, the user can acquire various information in life by using the device, and the user can perform effective health management unconsciously. For example, many methods have been proposed in which a sensor mounted on a hot water toilet seat, a toilet, or the like detects the amount or color of feces and uses the detection result for diagnosis.

Next, biological information (examination data) acquired in time series by the 1 st apparatus 2a and the like will be described with reference to fig. 2. As described above, in the present embodiment, 2 devices, i.e., the 1 st device 2a and the 2 nd device 2b, are assumed to acquire the user's examination data. Fig. 2 illustrates biometric information (examination data) acquired from 1 device out of 2 devices.

Fig. 2 is a graph created using the inspection data. The DB unit 8 records examination data arranged in time series for each patient ID, and fig. 2 shows the examination data in a graph. In fig. 2, the horizontal axis represents time T, and the vertical axis plots time-series examination data. On the vertical axis, the examination data, the biological data, the vital sign data, and the sample data are plotted, and any one of them is plotted based on the numerical value D of the examination output result of the examined device. The numerical value D is, for example, a value indicating the degree of redness of stool.

In addition, in fig. 2, it is assumed that the arrival date and time and the like are also systematically and automatically updated. The time of arrival may be plural, but may be simplified to avoid complication, for example, the time of arrival at the first visit of a specific clinical department. As described later, the example shown in fig. 2(a) is a case where the time-series data changes in a direction in which health deteriorates, and the user finally goes to the hospital. For the patient in the condition shown in fig. 2(a), if it is before time T1, it is possible to provide a result of reasoning about how long it will go to the hospital having what kind of clinical departments. In addition to this example, as shown in fig. 2(b), even when the patient is going to the hospital and vital sign data is obtained after other signs are detected, the information is recorded in the DB unit 8 in advance. However, there are also people who have not yet gone to the hospital at all but only have vital sign data.

As described above, fig. 2(a) is a case where it is assumed that the user visits the hospital later. The graph shown in fig. 2(a) shows a time-series change in examination data (device data) of a user who is not currently visiting the hospital. When a specific examination result (specific information) is obtained from the time-series examination data, it is possible to obtain information on whether or not the medical institution is present. Therefore, the health information that can grasp the health status of the patient can be guided until the patient gets worse to the hospital based on the time-series examination data. For example, in fig. 2(a), in the case of the examination data at time T1, it can be inferred that the medical institution is going to the time Tc when the time + Δ T elapses. That is, if the DB unit 8 stores examination data, medical institution information (hospital name, medical department, date and time information), and the like, it is possible to estimate a period until a medical institution receives a medical treatment.

Fig. 2(b) is a case of having gone to the hospital, which is a case of deteriorating in going to the hospital due to a reason other than treatment. The graph shown in fig. 2(b) is an example of a case where a person who has gone to the hospital due to a disease receives treatment in the hospital after specific information appears at times Tc1, Tc 2. In learning such a situation, the time-series inspection data shown in fig. 2(b) can be sufficiently used. This example is effective for guidance of the subject matter that "the person of the value is not usually able to treat himself". This is effective as information that can prevent further deterioration.

Fig. 2(c) is a situation without going to a hospital. In this case, the examination data D is lower than a prescribed value (shown by a broken line in the graph), and does not need to go to the hospital. In this case, the column of the arrival date and time is blank in the database of the DB unit 8 shown in fig. 2.

The DB unit 8 stores and records information of a hospital or a clinical department that has come, a disease name of a disease, information of a device, and the like. Therefore, even for a patient who does not take the apparatus into consideration, the optimum facility can be recommended. The database may hold the relationship between the type of information to be acquired (occult blood test information of toilet), the hospital, and the equipment Mod, or the time-series data of different patients may be managed by different databases. Further, the information corresponding to the database recorded in the DB unit 8 may be obtained by searching a plurality of DBs and sorting the search results.

Fig. 2 shows time-series information for each patient recorded in the DB unit 8 in a graph, in which the horizontal axis represents time and the vertical axis represents a value obtained by digitizing acquired information. Therefore, the information becomes two-dimensional visual information. Since the two-dimensional map is obtained, the following two effects can be said. First, since this is a diagram, it is possible to perform processing in the same manner as image determination, and it is possible to easily implement inference by simply leaving a generic and easily constructed AI chip or system such as an inference model for image recognition. Further, since the horizontal axis represents time, information on temporal changes in the body information can be effectively used, and prediction and the like can be simplified. Further, information relating to a characteristic of temporal change unique to a living body such as fluctuation or frequency can be added.

For example, the user can easily consider information on the time of sleep, the time of getting up, the morning day and night, before meals, after meals, and before and after bathing. In addition, there is a study that a proper fluctuation of heartbeat or respiration is more relaxed and favorable for health.

In fig. 2, a certain period of the period during which the history data is acquired corresponds to a specific period, and the inspection data in the specific period is extracted. The extracted inspection data as a time series is input to the inference engine 7, and the inference engine 7 outputs a suggestion by inference to provide the suggestion to the user. The specific time width may be a time width corresponding to some advice that can be given at a time point equivalent to the future after the end time point of the time width, or may be a width in which a plurality of pieces of information that are traced back with respect to the advice time point can be acquired. The specific time width may not be a strict time width determined by the specification, and a sufficient amount of data may be obtained. Since the time interval of each data is also important information, it is preferable that the data is non-discrete data obtained with a regular time width. However, even if the measurement time points of the data are discrete within the time width, it is effective if the time width is such that meaningful data is obtained by complementing the data by complementation. Or may be determined according to certain health and medical related information.

Further, the accuracy of prediction and the like can be switched according to an inference model by appropriately defining the horizontal axis width of the graph. If the time width is about 1 year, it is possible to predict the disease in the order of several months, and if the time width is 1 week, it is appropriate to predict the disease in the order of several days, and the appropriate width can be changed according to the characteristics of the disease. For example, when a disease that progresses slowly such as a tumor and a disease that cures or deteriorates rapidly such as an infectious disease such as influenza must be observed, appropriate time widths are different. That is, the information delivery apparatus according to the present embodiment includes a delivery information determination unit that extracts a change pattern of the user's examination data in a predetermined time width and determines delivery information to be delivered to the user in accordance with an inference model learned together with time information.

Next, a case where the user performs measurement in the 1 st device 2a and the 2 nd device 2b to acquire a plurality of pieces of inspection data will be described with reference to fig. 3. Fig. 3 graphically illustrates the inspection data obtained. In fig. 3, the horizontal axis T represents the measurement time, and the vertical axis represents the value of the inspection data (DA if the 1 st equipment and DB if the 2 nd equipment).

Fig. 3 illustrates that the following exists: even if the inspection data of the same item is for the same subject person, the same value is not output depending on the equipment or the setting/measurement environment. For example, when the threshold value for determining whether or not the health needs to be precisely checked is represented as DAR in the 1 st device and is represented as DBR in the 2 nd device, the measurement value exceeds the threshold value in the 1 st device and falls below the threshold value in the 2 nd device depending on the time from the results shown in the figure. That is, instead of determining that the time at T0 exceeds the threshold, the time T0 exceeding the threshold is traced back to search for T1, T2 … T4, and the inspection data, thereby making it possible to determine the time accurately.

However, since there is only one inspection history using the 1 st device 2a, it is preferable to refer to data stored in another device when more accurate determination is to be made. Then, by referring to data of a device (the 2 nd device 2b) capable of measuring the same measurement items as the 1 st device 2a, the amount of information is increased as shown in the graph 33, and the temporal change is accurately determined. For example, since bodily fluids such as excreta and blood vary depending on physical conditions, meal times, drinking or taking medicines, before and after bathing, before and after sleeping, and excess or deficiency (state of life), it is desirable to determine all of them without performing a single examination. In particular, in the case of a device for monitoring on a daily basis, various error factors are encountered, and therefore, it is desirable to perform such a countermeasure as described herein.

In addition, before and after a specific treatment such as an operation, vital sign data may be greatly changed due to a large physical burden. In addition, such patient examination data may change greatly with a certain time as a boundary, and when such a situation is determined and detected, it is preferable not to perform the correction as described in the present embodiment. In addition, by separately managing and determining the inspection information obtained in daily life after the time point, the value as the data for observation is also improved.

In order to reduce the error caused by the life condition described above, the determination is performed by adding data measured in various conditions, and is not performed only in accordance with a specific condition at that time (the data before and after the treatment such as the operation described above is not limited to this). That is, even if the examination data set of the subject person is acquired by the 1 st device 2a, it is preferable to be able to check the time-series change state, but the data amount is too small in the example shown in fig. 3. Then, the 2 nd inspection data set of the subject person is acquired as a time series by the 2 nd device (the 3 rd device or the 4 th device may be used) capable of complementing the 1 st inspection data set, thereby complementing the data, and making it possible to determine with high reliability. The 2 nd device includes a 2 nd inspection data acquisition unit, and records a history of the data by tracing back a predetermined time (up to time T4 in the example shown in fig. 3). When the delivery information determining section determines the delivery information to be provided to the target person using the 1 st inspection data group and the 2 nd inspection data group, the reliability becomes high. The 1 st examination data group and the 2 nd examination data group can provide information that reduces the influence of the special life condition as described above by complementing the examination time and the examination item with each other. Here, the specific time is traced back to T4, but in the above-described observation, there may be a design traced back to the post-operation or the like.

More specifically, an example in which prevention of a disease in the digestive system is considered will be described. As a method for preventing such a disease, a method such as fecal analysis is known. In the example of fig. 3, it is assumed that the subject person (user) unknowingly performs the stool analysis using a toilet in which the stool analysis is performed at the time of excretion. The chart 31 is created based on the examination data acquired by the 1 st facility 2a in the home toilet, for example, and the chart 32 is created based on the examination data acquired by the 2 nd facility 2b in the office toilet, for example.

Thus, if the data of two toilets can be used in a private or public environment as a life model, many inspection opportunities will be presented, and abundant information can be acquired. The opportunity to correct the situation where high data or the like appears due to tension in the workplace is increased, or to determine a tendency or the like that is not noticed outside the workplace is increased.

Even if the same user performs the examination in the same examination item, measurement errors, accuracy, and the like vary depending on the equipment used, the management system, and the like. Therefore, it is not always correct to record 2 pieces of inspection data on the same graph as shown in the graph 33. For example, the 1 st device 2a used in a home may be a simple sensor, and the 2 nd device 2b used in a workplace may be a high-performance sensor that emphasizes health management of a practitioner. In such a situation, although the 2 pieces of inspection data have similar values, they may not be easily compared depending on the measurement method or the sensor. In another viewpoint, when the 1 st facility and the 2 nd facility are assumed to be disposed in a station or a toilet of a public facility, it is difficult to manage the temperature and humidity in this case, and since many users are used, it is easy to generate an error due to the environment such as temperature, or an error factor such as deterioration of parts or contamination, and it is difficult to perform numerical analysis at the same level (scale). However, in order to collect health management information of a specific person, it is desirable to increase the amount of information by comprehensively using data acquired by various devices, thereby improving the value of the information. In addition, when the toilet facility is frequently updated, data can be processed equally before and after the facility change. This is preferably also taken into account for processing the data.

In the present embodiment, the inspection data acquired by the 1 st equipment 2a and the 2 nd equipment 2b are not simply arranged as shown in the graph 33, but as shown in the graph 34, 2 pieces of inspection data are corrected and calculated assuming that the same equipment has the same numerical characteristics. In fig. 3, a graph 34 is shown as a graph in which a predetermined number of cases in which the output from the 1 st device tends to appear low as in graph 33 are increased or a gain is applied to enhance the cases. That is, the inspection data is set as a database in which the ID of the device and the inspection result are recorded together, and the data of the same device is subjected to offset correction (addition/subtraction) and/or gain correction (multiplication/division) uniformly. By this correction calculation, the vertical movement pattern of the data transition that changes with the acquisition time obtained in the plurality of devices can be compared with other data in the same manner.

However, when the pattern differs even if the pattern is corrected, such as when the pattern of increase and decrease of the examination data is reversed depending on the degree of health, the measured values are recorded in advance so as to be analyzable so as not to make an erroneous comparison. Further, information such as the type of the device used for measurement, the model thereof, and the sensor value may be used. If this information is present, it is possible to compare the numerical change patterns from other devices that acquire similar biometric information. In general, since the biological information is rarely changed on a minute scale, an error in clock information (which determines the accuracy of the horizontal axis of the graph) of each device may be as accurate as a minute unit.

With the above-described consideration method, the health-related numerical values in the appropriate time range for the change in physical condition, the onset or the deterioration of the disease are obtained in a large amount, and therefore the information providing unit 1c can make an early suggestion to the user based on these pieces of information. When inference is made using discrete time-series data of a specific device, data is formed while correcting information that complements the interval of the discrete time, thereby enriching the data.

Next, an example of an operation of transmitting the inspection result in the information delivery system will be described with reference to a flowchart shown in fig. 4. The CPU in the control unit 1 controls the entire information transmission system in accordance with a program stored in the memory, and executes the flow.

In the example shown in fig. 4, it is assumed that the 1 st device 2a is a wearable device and the 2 nd device 2b is a dedicated device. Each is a device for acquiring health-related information of a subject (a specific user). The wearable device has low measurement accuracy, but the wearable device can frequently perform measurement and collect a large amount of information because the wearable device is worn close to the skin on a daily basis. On the other hand, although the measurement accuracy is high, the dedicated device cannot perform frequent measurement as compared with the wearable device. By acquiring the inspection data by the user using a plurality of devices, it is possible to complement each other's drawbacks, improve the accuracy of the data, and increase the number of data. That is, the respective advantages of a plurality of devices having different characteristics can be enhanced, and health management closely linked to each daily life can be realized.

When the flow of transmission of the examination result shown in fig. 4 is started, the mobile terminal acquires the health-related value (S101). Here, the wearable type 1 st device 2a acquires a health management value (test data) such as blood pressure on a daily basis. The inspection data acquired by the 1 st device 2a is transmitted to the control unit 1 when abnormal, and the process proceeds to step S103. If there is no problem with the loss of energy or time consumed for communication or inference, the check data acquired in step S101 may be transmitted to the control unit 1. As described later, since the inspection data of the 1 st device 2a is used when the matching adjustment is performed on the history data in step S107, the 1 st device 2a transmits the inspection data to the control unit 1 at predetermined time intervals.

When the terminal acquires the health-related value, the terminal searches for the result of the examination of the same person (S103). Here, the ID determination unit 1b of the control unit 1 searches the DB unit 8 for the result of the examination by the same person as the user measured in step S101. In a terminal owned by an individual, a dedicated sensor is a simple sensor, or an error is easily included due to a restriction on movement or handling of the individual. Therefore, it is desired to check and verify whether or not the health-related value acquired by the terminal includes no error, using the result of the health-related value acquired by the dedicated device. As described above, this step may be skipped also in the case where the check data is not abnormal and the check data is not transmitted.

Next, the dedicated device acquires the corresponding data (S105). Here, the 2 nd device 2b of the dedicated device type acquires the health management value (examination data) of the user. When the 2 nd device 2b acquires the inspection data, the inspection data is transmitted to the control unit 1. The dedicated device is a device processed and corrected by a specific professional institution or expert, and is often installed and used in a stable environment, and it is expected that a highly reliable result is obtained from the result of a terminal for personal use. In contrast, a use method can be employed in which, for example, the result obtained in step S101 is corrected based on data obtained by the dedicated device and processed.

Next, the history of the inspection data acquired in steps S101 and S105 is adjusted, and inference is performed using the adjusted history data (S107). As described with reference to fig. 3, even if the 1 st device 2a and the 2 nd device 2b are the same, the values of the inspection data may be shifted due to device errors and the like. However, if the same person is used, the trend of the inspection data acquired in time series is the same even if the apparatuses are different. Then, the control unit 1 performs a correction operation on the inspection data acquired by the 1 st equipment 2a and the 2 nd equipment 2b to generate inspection data (history data) as if the inspection data were acquired in the same equipment as a time series. The inference engine 7 inputs the history data and performs inference for outputting a recommendation for a change in physical condition, onset, or deterioration of a disease.

After the inference is made, the inference result is then displayed (S109). Here, the control unit 1 transmits a suggestion based on the inference result to the terminal 4 held by the user, and the terminal 4 displays the suggestion. In this case, it is possible to learn what kind of label information the inference model has, and to change the advice in various ways, and for example, if the label information includes the type of clinical department of a hospital to be visited and prescription information, it is possible to present such a result as an output of inference. Further, not only the inference results, but also an internet search may be performed on a database or the like based on the inference results, and additional notice or the like may be presented based on the search results. Further, it is also possible to determine what trend of the inspection data is in what situation based on the history of the inspection data of the user, the history of the held mobile terminal, and the like, and the determination may be made by inference, or may be made based on a rule. That is, if a person who is considered to have a sign of hypertension determines the situation of a user who outputs a high value in the history data, it is possible to analyze and present what is the trend of hypertension in some cases, and depending on the situation, it is possible to present an influence of pressure or the like in the workplace.

Even if only the date and time at which the high numerical value appears is presented, the user can observe the date and time, and can analogize the trend of the health of the user according to what environment, season, and what time period of the day, etc. The user starts actions such as taking a rest, taking a medicine, going to a hospital, etc. based on the judgment, and therefore, measures can be taken in advance before conscious and people who continue a healthy life can be increased. Further, if the user knows an environment in which the numerical value of the inspection data tends to deteriorate, when the user comes to the environment, the user can increase the clue information for specifying the cause by performing feedback control such as control for causing the portable device or the inspection device to acquire various data.

In the flow of transmission of the examination result according to the present embodiment, it is considered that the result obtained by the wearable device (S1) having a large variation factor can be corrected by almost ignoring errors in the dedicated device or the health diagnosis result. Therefore, it is possible to determine whether the health data obtained daily is different from or the same as the examination in the dedicated device.

As a result, the numerical value is occasionally good when the measurement is performed by the dedicated device, and the determination can be accurately performed when the numerical value measured in the life scene shows a worse result. In this case, for example, it is possible to take care of health advice such as attention after a meal or attention when getting up. The health advice may be presented by a device or the like that is easily accessible to the user, such as the terminal 4 of the user. Recently, technologies capable of being displayed on a user's television and transmitting information to a specific individual in addition to a portable terminal such as an AI speaker and a health management washbasin have been widely provided, and therefore these technologies can also be used.

In describing the flow of fig. 4, the description has been mainly given of a device provided with a dedicated sensor such as a sphygmomanometer, but the present invention is not limited to the blood pressure, and can be applied to information such as pulse rate and heartbeat. These can be supplemented by a general or widely known technique that can be mounted on a mobile terminal other than those for special applications such as an imaging element and an acceleration sensor. The 2 devices need not be devices that measure the same item (e.g., blood pressure and blood pressure), but one may be blood pressure and one may be pulse. In the case where there is a correlation such as what blood pressure is when there is any heart rate or in the case where there is no correlation, if the state where the heart rate is high continues, the risk of the onset of heart disease in the hypertensive patient increases, and therefore, it is often helpful for comprehensive judgment. Since various devices can acquire various health relationship data, the monitoring device or the data to be noticed may be changed by the user, and a recommendation as to which hospital to go or the like may be made according to the symptoms.

In this way, the information providing unit 1c may customize the information for each user. Specifically, it is considered to provide information and the like related to an appropriate clinic near the residence of the user. In addition, it is also possible to determine which values to monitor by a medical staff of a facility which visits frequently, and to manage data by a system provided in the facility. If the medical facility has handled a sufficient number of cases, it can be diagnosed from the person's health information similarly to the person's health information that trends or the like with the same disease are respectively performed. Therefore, the DB unit 8 may be provided in a server in a hospital to store data. In this case, it is possible to perform inference, investigation, or suggestion that automatically reflects the environment, eating habits, and the like specific to the area.

Next, another example of the transmission operation of the inspection result in the information delivery system will be described with reference to a flowchart shown in fig. 5. This flow is executed mainly by the CPU in the control unit 1 controlling the entire information delivery system in accordance with a program stored in the memory. The flow shown in fig. 5 represents a case where the functions of the search in the DB unit 8 shown in fig. 1, the inference engine 7, and the like are used alone. There may be a case where either of the functions is used or a case where both of the functions are used in a superimposed manner, but the simplest example is shown here.

In explaining the flow of fig. 5, the description will be given assuming that an image sensor, a magnified image determiner such as a microscope, a sensor for detecting reflection of special light, etc., a crystalline nanowire array, an olfactory sensor to which a change in electrical characteristics such as a molecular film is applied, a gas component sensor, etc., are disposed in a toilet bowl as the 1 st device 2a, the 2 nd device 2b, and the third device 3, and the characteristics of excrement of a user can be confirmed.

When the flow of transmission of the inspection result shown in fig. 5 is started, first, determination is performed based on the sensor output result for each ID (S1). Here, there are a case where the control unit 1 obtains the output of the 1 st device 2a and the like through the communication control unit 1a and a case where the control unit 1 receives data transmitted by the 1 st device 2a and the like through the communication control unit 1 a. Further, a method is assumed in which the control unit 1 collects data recorded by the 1 st device 2a and the like at a specific time by the communication control unit 1 a. At this time, the determination of the inspection result is made based on the sensor output for each ID attached to the sensor output result. The sensor may be a color sensor, a shape sensor, a hardness sensor, an olfactory sensor (including reaction judgment of nematodes or animals), a gas component sensor, or a color change detection sensor when a specific reagent is added, and shape judgment based on an enlarged observation image may be performed based on an output of the image sensor.

When the characteristics of the excrement of the user are confirmed, for example, the feces having occult blood can be determined by a color sensor. The amount, shape, hardness, and the like of excretion may be determined by an image sensor or a color sensor, or a method of measuring a color distribution or the like by special dyeing may be employed. Alternatively, the composition may be detected using an image obtained by enlarging the object, and the result of the incubation may be determined at a specific time. For example, when blood mixed in feces is increased, the red color of red blood cells becomes conspicuous, and when it is quantified, a difference from the health condition can be known. In step S1, they are detected.

In step S1, the control unit 1 determines whether or not the specific information can be obtained after determining the sensor output result by a determination using a specific program or the like (S3). Here, based on the determination result in step S1, it is determined whether or not specific information associated with the disease, such as a characteristic such as a numerical value that is different from the health state, is detected.

If the specific information cannot be acquired as a result of the determination at step S3, the process returns to step S1. On the other hand, if the specific information can be acquired as a result of the determination in step S3, it is determined whether or not there is a passing inference model (S5). Here, based on the specific information acquired in step S3, it is determined whether or not a database in which precise examination and the like relating to the specific disease can be performed is stored and whether or not an inference model capable of inferring future progress using the data is set in the inference engine 7.

If the inference model is not passed as a result of the determination in step S5, a specification of inference is created (S13). In step S5, if it is determined that no database is stored, the control unit 1 requests the DB unit 8 to construct a database. By constructing a database capable of searching for a specific disease in advance, it is possible to quickly construct a system as users increase even for devices used for the first time. Further, if a system is simply constructed in which a person with a disease or a person without a disease transmits the meaning together with examination data, it is possible to determine whether or not a disease is to be caused based on the acquired data. Further, it is desirable to be able to make more accurate inferences than such simple predictions, and therefore, to obtain such inferences for purposes, first, a specification of inferences is made in this step. The detailed operation of creating the inference specification will be described later with reference to fig. 7.

After creating the specification of inference in step S13, creation of the specification of inference model is requested (S15). Here, the specification of the created inference model is transmitted to the learning unit 5 by the learning delegation unit 6. The learning unit 5 generates an inference model according to the specification. The control unit 1 receives the generated inference model through the learning delegation unit 6. After the inference model creation request is made, the process returns to step S1. The detailed operation of creating the inference model will be described later with reference to fig. 8.

If the determination result in step S5 is yes, that is, if there is a database for search and an inference model, the "history search" method is determined (S7). If the determination result in step S5 is yes, the database for search is present, and in this case, the information related to the specific disease determined in step S3 is searched from the data recorded in the DB unit 8. In this step, the method of history retrieval is decided, i.e. how to retrieve from the database the facilities for further examination for the user's specific affection. For example, if it is a disease of the digestive organ system, examination data of the excreta system is mainly retrieved. In addition, if the disease is a first-aid disease, a short-term history may be searched, and if the disease is a chronic disease, a long-term history may be searched. In this case, if the period is too long and there is a large amount of data, the data may be thinned.

It is difficult to predict the future from the previous pattern on the time axis using only 1 kind of data 1 time. Then, in step S7, history data in a specific time range traced back to the past on the time axis of the user' S health data is retrieved and used. Since the search result includes time information, future prediction can be performed. The specific time width varies depending on the disease. For example, if it is a recent future expectation that a disease state will change rapidly, recent past change data is important, but if it is a disease that gradually worsens like a lifestyle disease, a long-span history becomes important. Therefore, the time range of history acquisition may be changed according to the disease to be examined.

After the history search method is determined, one of the history data is matched with the other history data, and inference is performed (S9). In the present embodiment, the examination data of the user is acquired by two devices, i.e., the 1 st device 2a and the 2 nd device 2 b. As described with reference to fig. 3, the levels of the inspection results of the 1 st device 2a and the 2 nd device 2b do not match. Then, in step S9, the control unit 1 performs a correction calculation on the inspection data so that one level (the level of the inspection data of the 1 st device 2a) matches the other level (the level of the inspection data of the 2 nd device 2 b). In this case, the control unit 1 may match one history data with the other history data.

After the history data is subjected to the matching adjustment, the inference engine 7 performs inference for outputting a recommendation for a change in physical condition, onset of a disease, or deterioration using the history data in step S9. The detailed operation of "performing matching adjustment and inference on the history data" in step S7 will be described later with reference to fig. 6.

After the history data is matched and adjusted and inferred, the inference result is displayed next (S11). Here, the control unit 1 transmits the inference result in step S9 to the terminal 4 of the user, and causes the display unit of the terminal 4 to display the inference result. This step S11 is a step of providing information for examination or medical assistance to the user or the person concerned who becomes the information source acquired in step S1, and it is assumed that a display or a warning appears on the terminal 4. When the inference result is displayed, the process returns to step S1.

In this way, in the flow of transmission of the examination result in the present embodiment, the control unit 1 obtains the detection results of the sensors from the 1 st equipment 2a and the 2 nd equipment 2b (S1), and determines whether or not there is specific information on the health state (disease) based on the detection results (S3). When the specific information is acquired, it is determined whether or not there is a database related to the specific information and whether or not there is an inference model (S5), and when there is a database, a search is performed for the database. Then, the data acquired from the 1 st device 2a is subjected to a correction operation so that the levels of the values acquired from the 1 st device 2a and the 2 nd device 2b match. After the 2 pieces of history data are subjected to matching adjustment by the correction operation, inference is performed using the history data (S9), and an inference result is displayed. Therefore, when a user can acquire inspection data by a plurality of devices, the output levels of the respective devices can be matched, and therefore, the inspection data can be enriched. It is possible to use abundant inspection data and perform highly accurate prediction and inference. As a result, the user can perform health check in daily life and can accept advice according to the health state.

When there is no inference model (S5 → no), a specification for making inference for inferring the specific information is created (S13), and generation of the inference model is requested to the learning unit 5 (by the learning requesting unit 6) (S15). Therefore, the inference model corresponding to the health state of the user can be sequentially added.

In the flowchart shown in fig. 5, for example, when the specific information is not acquired in step S3, the user' S details, actions, lifestyle habits, and the like may be determined. By acquiring these pieces of information in advance, appropriate information can be provided. In addition, information such as age, sex, and previous disease, information on residence, eating habits, and food, and the like are also effective as information. The information may be prepared by acquiring a questionnaire through the terminal 4, inputting and acquiring the information when the information determination device 2 is installed, inputting the information through the association check mechanism 9 when the user goes to a hospital, or the like, or by collecting information existing on a network through these devices or the devices.

In the flowchart shown in fig. 5, the DB search (S7) and the inference (S9) are processed independently as separate processes. However, the present invention is not limited to this, and these may be processed in a combined manner. For example, there is also a method of searching for a DB after inference is performed, and inference output to a device such as a held instrument may be performed using an inference model learned by including information in the DB including inspection apparatus information at the time of learning. In this case, a display of "having a good quality inspection apparatus in the clinic" can be performed.

Next, the operation of "performing matching adjustment and inference on history data" in step S9 in fig. 5 will be described with reference to the flowchart shown in fig. 6. As described above, in this flow, the 1 st device 2a and the 2 nd device 2b are used to extract the change pattern of the user's inspection data in a predetermined specific time width, and the correction calculation is performed so that the output levels of the respective devices match, thereby performing matching adjustment on the 2 pieces of history data. Based on the matching adjusted historical data, inferences are made about health-related recommendations. The control unit 1 performs this process in cooperation with the inference engine 7, the DB unit 8, and the like via the communication control unit 1 a.

When the flow shown in fig. 6 starts, time-series data is acquired (S21). Here, time-series data corresponding to the specific ID recorded in the DB unit 8 is acquired. The time width of the acquired time-series data is a specific time width, but if data of a specific time width cannot be acquired, the time-series data is set to be in an acquirable time range. This is because, if there is no specific time width, the data is determined only in a specific situation, and the reliability is deteriorated. The specific time width is different between a disease that progresses over time such as colorectal cancer and a disease that progresses over a short period of time such as influenza. The type of the examination data depends on learning of the inference model, but is expected to be a graph of a specific item used in learning, and it is preferable not to infer the weight and the blood pressure together, for example. Therefore, it is preferable to perform inference based on auxiliary information in consideration of data such as information on what kind of sensor of what kind of equipment is.

After the time-series data is acquired in step S21, it is determined whether time-series data of a specific time width can be acquired (S23). For example, if the condition of occult blood is detected, it is determined whether occult blood is obtained in a width of several months. That is, the specific time width differs depending on the associated disease.

If the data of the specific time width is not obtained as a result of the determination in step S23, no inference is made (S35). Even if there is no information of a specific time width, the inference can be made according to expected reliability, but the inference may be difficult. Then, in step S23, when it is determined that data of a specific time width is not obtained, no inference is made. However, there are cases where a dangerous situation can be detected clearly, and in this case, it is sufficient to output emergency information before inference.

That is, when the acquired data is a numerical value in which a problem is conspicuous, since the delay in the long-term expected time is not performed by inference, a warning display is performed when the change is conspicuous. By this measure, even if no inference is made in step S35, it is possible to prevent the measure from being impossible in an emergency, and a highly reliable system can be provided in which information is output after sufficient data is collected. That is, in the present embodiment, when the numerical value changes converging to a specific change, the change pattern of the user's test data is cut out in a predetermined time width, and the inference is performed according to the inference model learned together with the time information. After the processing of step S35, the flow ends, and the flow returns to the original flow.

If the data of the specific time width amount is acquired as a result of the determination in step S23, the data acquisition device information and the acquisition time information are then associated with each data (S25). Here, the control unit 1 associates the data acquired in step S21 with information indicating which device among the 1 st device 2a, the 2 nd device 2b, and the like has acquired and information on the acquisition time, and records the data in the DB unit 8. By associating these pieces of information with the acquired data, the data can be positioned in the graphs shown in fig. 2 and 3. If the number of devices to be evaluated is increased, the influence of errors of the respective devices can be reduced while increasing the number of data to be evaluated in time series. Further, if it is known which data (including time information) comes from which device, even if reliability of a certain device is lowered by some factor at a specific time, it is possible to design to use only data before that time and to make a determination based on the data thus used. In addition, when it is possible to determine which device the user uses frequently, a usage method in which determination is mainly made only by the device and the result of the inspection using another device as needed is reflected can be adopted.

Next, the time-series data of the different devices is increased or decreased (S27). As described with reference to fig. 3, when the user's time-series inspection data is acquired by a plurality of devices, there are errors, differences in characteristics, and the like of the respective devices, and therefore, the plurality of time-series inspection data cannot be plotted on the same graph (see graph 33 in fig. 3). However, since the inspection data is time-series data of the same person, the trend of the change pattern of the data is the same. Therefore, by performing the correction operation on the plurality of time-series inspection data, the plurality of time-series inspection data can be plotted on the same graph. As the correction operation, addition and subtraction may be performed on each data based on a difference of average values of 2 time-series data or the like, and multiplication and division may be performed. As a result of this correction operation, time-series data corrected in accordance with different apparatuses is obtained. In addition, when certain specific data is important and other specific data is not important, a difference may be provided in the reflection of the information by weighting the data change or the like.

In step S27, the time-series data is added or subtracted, and then the added time-series data is collectively input to the inference model (S29). In step S27, since time-series data is generated for each device, the control unit 1 inputs the time-series data to the inference engine 7. In this case, the reliability may be determined to be low because the logic is not logical in a specific inference model that is learned using, as training data, increase/decrease information in the same time range as the specific time, which is an inference including an error possessed by each device. Then, the reliability of the inference is calculated while the correction of the different device is performed, and a result with high reliability is set as the inference result (S31). Here, the constants of the specific four arithmetic operations are changed little by little for time-series data of different devices and processed. This process increases the reliability in a situation where the error is corrected, and therefore, accurate inference can be realized.

For example, the control unit 1 may determine whether or not the devices detect the same biometric information based on the device ID of each piece of information, and then perform the process of step S31. Thus, since the increase/decrease relationship of data based on the change in health is secured, it is only necessary to reduce sensitivity, environmental errors, and the like. However, if the items are the same items of examination such as the increase/decrease relationship of data based on the change in health, the items may be uniformly processed. This is because, when the processing is performed uniformly, if the number of data is a disease state that is effective as the density in time or the range in time, more effective information can be obtained. In this case, the control unit 1 may be provided with a step of determining whether or not the data can be processed equally, or determining the data to be processed collectively assuming a specific disease, as long as the data has information on what kind of examination item the data has.

That is, if the information output from the device is not only the data of the determination result but also some information of the time (date and time) of the examination, the information corresponding to the individual to be the subject, the information of the examination content, the information specific to the device, the information of the device type, and the like in a specific format, the data can be corrected or selected by using the information. In addition, not only correction but also weighting can be performed. The data acquired by a device with low reliability may be designed to be weight-reduced and not to be processed as much as other devices. Further, if there are many people who use the same plurality of devices, for example, when trying to distinguish all colors as time-series data and arrange them on a graph, even if not information of a single device but information of a plurality of devices is mixed, people in the same health state tend to be the same.

That is, if a graph showing the transition of information is created so that information from a specific device acquired by a specific person at a specific time can be identified on a common time axis by color separation or the like, the trend of the health state of the person can be grasped and transmitted. The determination may be made by adding various corrections and the like that have been described on the graph. Recognition based on color separation or the like is a design that is easily understood by humans even when viewed by eyes, but in addition to this, the shape of a dot of data drawn and shown on the drawn dot may be changed so that device recognition is possible, and even if information added to the data of the dot is displayed or read, the same effect can be obtained.

After the inference is performed in step S31, an inference result is obtained (S33). Here, when the inference is performed in step S31, the control unit 1 sets the inference output with the highest numerical value of reliability as the inference result. Health advice that effectively utilizes information acquired in various life situations can be realized.

In this way, in the flow of matching and adjusting and inferring the history data shown in fig. 6, time-series data is acquired (S21), and when time-series data of a specific time width amount can be acquired (yes in S23), time-series data of different devices are corrected so as to have the same level, and inference is made using the history data subjected to the correction (S29). Then, the reliability of the inference is determined while performing the correction for each device, and the inference output with high reliability is set as the inference result (S31). Therefore, a highly accurate inference result can be obtained using the inspection data of the plurality of devices. Further, since the inference can be performed using time-series data of a specific time width, the inference can be performed with high accuracy.

In the present embodiment, when the numerical value changes so as to converge to a specific change, the change pattern of the test data of the subject person is cut out in a predetermined time width, and the inference is performed in accordance with the inference model learned together with the time information. That is, when the predetermined criterion is not satisfied (when the data of the predetermined time width cannot be acquired (no in S23)), the change pattern of the test data of the target person is not cut at the predetermined time width and the inference is performed according to the inference model learned together with the time information.

Next, the operation of "creation of inference specification" in step S13 in fig. 5 will be described with reference to the flowchart shown in fig. 7. The subroutine of creating the inference specification is to create a specification for requesting the learning unit 5 to create an inference model when it is determined in step S3 that the specific information is acquired and the inference engine 7 does not set an inference model for performing inference based on the specific information.

When the flow of creating the inference specification is started, first, the related disease is determined based on the specific information (S41). The associated lesion is determined based on the specific information determined in step S3 (fig. 5). For example, since the biological information related to what kind of disease is determined based on the urine test result or the feces test result of the user, if a table or the like indicating the relationship between the test item and the related disease is recorded in the DB unit 8 or the like, the related disease can be determined based on the recorded table.

When the associated disease is known, the patient with the associated disease is then determined (S43). The DB unit 8 records and stores daily health information (biological information, examination data) of a large number of patients. Then, the control unit 1 determines (searches) a patient other than the user who has the related disease determined in step S41.

Next, it is determined whether or not there is a history of the health information of the patient (S45). Here, the control unit 1 determines whether or not the history of the health information of the patient determined in step S43 is recorded in the DB unit 8. If the determination result indicates that sufficient data (history of health information) is not stored, it is assumed that no advice information is present without making an inference or an inference request (S51). The flow is ended and the original flow is returned.

On the other hand, if the determination result in step S45 is that there is a history of health information, the history data is input for each time width of the associated lesion and the lesion is output (S47). Here, the control unit 1 searches for health information of a patient whose disease has been diagnosed recorded in the DB unit 8, and determines whether or not there is a period or amount in which the health information can be used as training data. If the determination is satisfied, the historical data is extracted within the time-series data of the patient over a time width required for determining the determined associated disease, and the time-series data is used as training data. The time-series data is information on a patient who has already been diagnosed with a disease, and includes biological data measured by an instructor having expert knowledge using a relatively accurate measurement device. Then, the control unit 1 corrects substitute data obtained by a home appliance or a portable terminal, which is obtained in a daily period, with the biological data (series of data) as a reference, and sets the time-series data as training data. The control unit 1 inputs the training data to the inference engine 7, and acquires and outputs the disease information.

The biological data (series of data) described above differs in pattern if there is no information such as which time point of the disease is equivalent. For the inference for sub-health, a pattern traced back to the time point may be used based on the time of the initial hospital arrival or the time of the initial disease diagnosis. That is, data of a patient who goes to a hospital after a predetermined period shown in fig. 2(a) and data of a person who may not go to a hospital shown in fig. 2(c) are collected in large quantities, and a difference between them is determined by learning to create an inference model. If inference is made using this inference model, the resulting data can be used to decide which patterns are similar, and to decide whether or not to become a disease, etc. In the data in the case of being taken after the start of hospital visit as shown in fig. 2(b), the pattern may vary depending on the treatment. However, since there is a demand for reasoning in the future by including the effect of the medicine, there is also a method of creating and learning training data that is known about the time point when the patient starts to go to the hospital.

When the inference is performed using an inference model created as training data by correcting a time series data set in order to improve the reliability of the inference, a highly accurate prediction can be performed by a method of correcting the time series data set and inputting the corrected data set and using an inference result with high reliability. In addition, if there is biological information that changes over time even if it has no direct relationship with the disease, the other reference information is set as different data as training data.

After the specification for the inference model creation is created in step S47, the specification is further set to a specification in which a recommendation for a disease and a current sub-health level are collectively output (S49). Here, the inference engine 7 creates a specification of an inference model capable of outputting a recommendation for a disease. At this time, a suggestion of which stage the lesion is in may be output, and information of which time the user is currently before going to the hospital may be output. After the specification for creating the inference model is created in steps S47 and S49, the flow is terminated and the original flow is returned to.

The learning of the creation target of the specification in the flow of fig. 7 can be applied to the fields such as the relationship between the gust and the labor time. If the frequency of labor is learned at a timing traced back on the basis of birth, an inference model can be generated that can suggest how long later to go to the hospital or call for a parturient.

Next, the learning operation performed by the learning unit 5 when the "inference model creation request" is performed in step S15 of fig. 5 will be described with reference to the flowchart shown in fig. 8. Here, an inference model is generated that can obtain a set output using the history data of the patient with the associated disease determined in fig. 7 as training data. The subroutine of the inference model creation is mainly executed by the input/output modeling unit 5a in the learning unit 5.

Training data is typically produced by labeling specific data with specific labels. In this flow, health-related information (examination results, date and time of arrival at a hospital, advice, and the like) of the person is given to a data (acquisition information) group acquired at a plurality of times in association with the same subject person, and the data is used as one piece of training data. By preparing the training data of a plurality of subjects, it is possible to infer what health information the transition pattern of the time-series data corresponds to. The raw data to be the training data may be recorded in the form of a file, or may be a metadata set required to be able to be recorded in association with each other. The metadata may also exist for annotation. Further, since selection is made as necessary when creating the inference model, information such as an ID for specifying the created inference model may be recorded as metadata in a file used as a basis for the inference model. By these processes, the AI can be prevented from being blackened.

When the flow of inference model creation starts, first, input/output is set (S61). Here, the learning unit 5 sets the input and output of the inference model based on the specification transmitted from the control unit 1 through the learning delegation unit 6. That is, what (what information) is input to the inference model, what (what information) is inferred and output, and the like are set. The number of intermediate layers of the neural network is set, and the weight of each intermediate layer is set to an initial value. In this step, a so-called "requirement specification" of the inference model thus created is set.

Next, training data is input to create a model (S63). The control unit 1 creates training data from the data recorded in the DB unit 8 and transmits the training data to the learning unit 5 (see S47 and S49 in fig. 7), and thus sequentially inputs the training data to the input unit of the input/output modeling unit 5 a. Since the training data is a set of input and output, the inference model is created by determining the weight of each intermediate layer of the neural network so as to be output in accordance with the input.

After all the training data are input in step S63, it is next determined whether or not a model has been created with high reliability (S65). Here, it is determined whether or not the value indicating the reliability of the inference model generated in step S63 is higher than a predetermined value.

If the determination result in step S65 is that a highly reliable inference model cannot be created, relearning is performed (S69). Since the reliability is low, the parent set of the training data is changed, and the process returns to step S63 to create the inference model again. When the reliability does not reach the predetermined value even after the relearning is performed a predetermined number of times, the generation of the inference model is ended, and this is sent to the control unit 1. In the case of relearning, metadata indicating that a data group or a file is not used as training data may be additionally recorded. It is possible to prevent a situation in which a data group or a file of poor quality is used for learning and cannot be smoothly performed.

On the other hand, if the determination result in step S65 is that a highly reliable model can be generated, the model is set as an inference model (S67). The learning unit 5 transmits the inference model generated here to the control unit 1. When the specific information that is the basis of the inference specification created in step S13 is acquired, the control unit 1 can set the received inference model in the inference engine 7 and perform inference. After the inference model is created, the flow is ended and the original flow is returned.

In this way, in the flow of creating the inference model, the learning unit 5 to which the training data is supplied learns to create the inference model. In this flow, training data is selected and data to be included in the training data is selected and relearning is repeated (see S69) until highly reliable inference can be made (yes at S65). By this learning, it is possible to create an inference model in which the change characteristics of the health numerical change pattern in the sub-health stage are compared with the pattern of a disease.

In the process of selecting training data for relearning, if there is data that is not related to a specific disease, the control unit 1 may consider that the reliability is not high, and may exclude the independent data or the data group. If the control unit 1 tracks the specific biological data excluded in such a process, the device that has output the data, the environment in which the data is obtained, or the like, it is possible to specify the data, the device, or the environment that are not suitable for inference, and it is also possible to consider recording the specific data, the device, or the like and exclude the data, the device, or the like when creating the inference model next time or later.

If the unavailable data information exists, it is possible to screen the biological information acquisition device, data, or environment to be used when there is a sign of a specific disease. For example, although a stool examination history is important for detecting signs of colorectal cancer, it is preferable to avoid processing the heart rate and the like, which are information on different organs, in the same manner, from the viewpoint of increase in data amount and complexity of calculation. However, if the same type of biological information is used, it is often preferable to increase the number of data items or perform tracking, even if differences in the devices and differences in the measurement methods are taken into consideration.

On the other hand, even if training data is formed using the observation and stool result that has not changed and measured by another device for many years and the recent observation and stool result that has changed rapidly, learning with a great effect is not possible. If the number of data obtained in a unit time (see, for example, S3 in fig. 4) is insufficient or if the precision is poor and the data thus eliminated cannot be used, among the data of the observation and defecation results obtained when the latest specific device is used, the data of the observation and defecation results obtained when the device is not used are supplemented with the data of the observation and defecation results obtained when the device is used, the data information amount of the observation and defecation results increases, and therefore, there is a high possibility that the data becomes effective training data. That is, if the amount of information input to the inference model is appropriate, the reliability of the inference result also becomes high.

Next, a modified example of the operation of performing matching adjustment on the history data and performing inference (see fig. 6) will be described with reference to fig. 9 and 10. The example of fig. 3 described above assumes that the data amount is insufficient and appropriate inference cannot be performed only with 1 device, and the inference result becomes accurate by supplementing information using a plurality of devices. However, as shown in fig. 9, when the data history about the specific person provided in each device is sufficient, as shown in a graph 34 in fig. 3, it is also possible to perform inference in an inference model for each device without performing information amount supplementation per unit time and comprehensively determine the result. Further, since it is possible to realize accurate determination by adding not only the inspection time but also information from the device that outputs different information and performing determination based on the added information, it is also possible to improve the reliability of the inference result by supplementing the information itself of the device that supplements the inspection data of different inspection items.

When a data group, a file, or the like is put into the inference model, metadata indicating a data group or a file created for the purpose of inferring by what kind of inference model can be associated in advance, and an optimal inference model can be specified by the metadata. For example, when health-related information to be inferred is specialized, it is possible to separately prepare an inference model for colon cancer and an inference model for hemorrhoid. With such a design, information such as hemorrhoids can be output to a user who is mainly concerned about colorectal cancer and the like without waste. Further, if information for making the result of inference metadata is associated with a data set or a file used for inference in advance, the information becomes effective information when searching for what kind of data set or file what kind of case is.

In this case, the device that measures the training data used for learning differs from the device that obtains the data input to the inference model for each device. For example, a graph 91 shown in fig. 9 shows a change in the inspection data acquired by the 1 st equipment 2a as a time series, and a graph 92 shows a change in the inspection data acquired by the 2 nd equipment 2b as a time series.

Since the difference is caused depending on the equipment, the correction value is changed little by each correction input 91a, 92a for each inspection data, and the correction operation is performed. Then, the corrected inspection data of the 1 st device 2a is input to the inference engine 7a for the 1 st device, and the corrected inspection data of the 2 nd device 2b is input to the inference engine 7b for the 2 nd device. In this way, each corrected data is input to the corresponding inference model.

When the correction value is changed little by little, the reliability of the inference output also changes little by little, and therefore the reliability of the inference output employed for each device becomes an appropriate inference result. When the inference results are decided in accordance with the respective devices, the results that comprehensively reflect the respective inference results are taken as final outputs. As a comprehensive judgment, an inference result with high reliability can be selected, and if the reliability is not greatly different, an intermediate judgment between the two results can be adopted, or a judgment containing the two results can be adopted. As described above, in the present modification, the result of inference using a plurality of time-series data corresponding to a plurality of inference models is comprehensively determined and used as the inference result, and therefore, a highly accurate advice display method can be provided.

Next, the operation of a modified example of performing matching adjustment and inference on history data will be described with reference to a flowchart shown in fig. 10. As in the case of fig. 6, the control unit 1 performs this process in cooperation with the inference engine 7, the DB unit 8, and the like via the communication control unit 1 a.

When the operation of the flowchart shown in fig. 10 is started, first, inference is performed while correcting the numerical value of the history of the 1 st device (S71). Here, as described with reference to the graph 91 and the correction input 91a of fig. 9, the control unit 1 performs a correction operation by addition, subtraction, multiplication, division, or the like on the time-series inspection data acquired by the 1 st device 2 a. The control unit 1 inputs the corrected inspection data as a time series to the inference engine 7a, and causes it to perform inference.

Next, the result of the correction value for which the reliability is appropriate is employed (S73). As described above, the control unit 1 calculates the reliability of the inference while changing the correction value of the correction operation little by little. In this step, the control unit 1 adopts the inference result when the reliability becomes the highest as the inference result when the history data of the 1 st device is used.

Next, the 2 nd device estimates the history value while correcting it (S75). Here, as described with reference to the graph 92 and the correction input 92a of fig. 9, the control unit 1 performs a correction operation by addition, subtraction, multiplication, division, or the like on the time-series inspection data acquired by the 2 nd device 2 b. The control unit 1 inputs the corrected inspection data as a time series to the inference engine 7b, and causes it to perform inference.

Next, the result of the reliability becoming an appropriate correction value is employed (S77). As described above, the control unit 1 calculates the reliability of the inference while changing the correction value of the correction operation little by little. In this step, the control unit 1 adopts the inference result when the reliability becomes the highest as the inference result when the history data of the 2 nd device is used.

In steps S71 to S77, after the inference results are decided for the 1 st device and the 2 nd device, respectively, then, if the employed results are similar, they are used for the suggestion (S79). Here, if the results employed in steps S73 and S77 are similar, the inference result is employed as a suggestion. In the case where the results employed are not similar, the determination may be made by reasoning which is likely to be correct, or the like. The determination may be performed comprehensively as shown in fig. 9.

In this way, in the modification example of performing matching adjustment and inference on the history data, the 1 st device inference engine and the 2 nd device inference engine each infer the inspection data outputted from the 1 st device and the 2 nd device as time series. In this inference, correction computation is performed on the inspection data of each time series, and the inference at the time of the highest reliability is adopted as the inference result for each device. And finally, comprehensively judging by using the reasoning results of the 2 devices.

In the present modification, 2 devices, that is, the 1 st device 2a and the 2 nd device 2b are used, but the present invention is not limited thereto, and the processing may be performed using 3 or more devices.

As described above, the information delivery system according to the embodiment of the present invention includes: a 1 st inspection data acquisition unit (ID determination unit 1b) that acquires a 1 st inspection data set of the subject person as a time series by a 1 st device 2 a; a 2 nd inspection data acquisition unit (ID determination unit 1b) for acquiring a 2 nd inspection data set of the subject person in time series by the 2 nd equipment 2 b; and a delivery information determining unit (information providing unit 1c) for determining delivery information to be provided to the subject person by using the 1 st inspection data group and the 2 nd inspection data group. The examination data sets are acquired from the 1 st device and the 2 nd device, and information to be provided to the subject person is generated based on the data. That is, since the inspection data is acquired from a plurality of devices, the number of data can be increased, and since the data can be acquired in various situations, information with higher accuracy can be generated. As described above, the information delivery system according to the embodiment of the present invention can grasp the accurate health state in consideration of the condition of the subject person, and can provide customized information such as advice corresponding to the health state.

In one embodiment of the present invention, the transfer information is determined according to an inference model learned according to a change pattern of the inspection data set acquired by the plurality of devices. That is, in the present embodiment, the inference model is generated using the inspection data sets acquired in time series (see, for example, fig. 8). The inspection data set acquired from the subject is input to the inference engine, and the inference result is obtained (see, for example, S107 in fig. 4 and S9 in fig. 5). By using the time-series test data set, it is possible to specify a disease represented by a change pattern of the test data, and to infer a future onset disease, its onset timing, and the like.

In one embodiment of the present invention, the 1 st inspection data group and the 2 nd inspection data group are corrected in accordance with the 1 st inspection data group acquired by the 1 st device and the 2 nd inspection data group acquired by the 2 nd device, the reliability when inference is performed using the corrected inspection data group as input is calculated, and the transmission information is determined in accordance with the reliability (for example, see fig. 3 and 4). Therefore, when the inspection data of a plurality of different devices is acquired, even if the output levels of the devices are different, the inspection data is corrected, and therefore, the transmission information with higher accuracy can be determined with a large amount of data.

In one embodiment of the present invention, the 1 st and 2 nd inspection data groups are corrected in accordance with the 1 st and 2 nd inspection data groups acquired by the 1 st and 2 nd apparatuses, the corrected inspection data groups are combined into 1 inspection data group, and the combined inspection data group is input to an inference model to perform inference, and delivery information is determined based on the inference result (for example, see S9 in fig. 3, 4, and 5). Therefore, the inspection data sets acquired by the plurality of devices are processed as if the inspection data sets were 1, and the number of data sets increases, so that the transmission information with higher accuracy can be determined.

In one embodiment of the present invention, the 1 st and 2 nd inspection data sets are corrected for the 1 st and 2 nd inspection data sets acquired by the 1 st and 2 nd apparatuses, the corrected inspection data sets are input to the inference model, the inference results by the inference models are comprehensively determined, and the delivery information is determined based on the determination results (see, for example, fig. 9 and 10). Therefore, it is possible to determine the transmission information with higher accuracy from the inspection data sets acquired by the plurality of devices.

In one embodiment of the present invention, the 1 st inspection data acquisition unit and the 2 nd inspection data acquisition unit determine whether the inspection data set is from the subject person or from a person other than the subject person (see the ID determination unit 1b in fig. 1), and in the case of the inspection data set from the subject person, acquire the inspection data set as the 1 st inspection data set or the 2 nd inspection data set.

When the inspection data from the 1 st equipment 2a and the 2 nd equipment 2b used by the subject person and the inspection data from the 3 rd equipment 3 used by a person other than the subject person are input, the control unit of the information delivery system can distinguish the inspection data of the subject person from the inspection data of the non-subject person. Therefore, the subject person can obtain the inspection data by using a plurality of devices, the inspection data is enriched, and the transmission information with higher precision can be obtained.

In the description of the present embodiment, the case where the results of analyzing the feces, collecting the feces, and the like by various sensors attached to the toilet are effectively used has been described a plurality of times as the 1 st facility 2a or the 2 nd facility 2b, but the present invention is not limited to this. The 1 st device 2a and the 2 nd device 2b may be any devices for acquiring health-related information of the subject, for example, vital sign information, sample information, and the like. In the simplest example, the present invention can be applied to face image information obtained from a portable terminal such as a smartphone, heartbeat information based on the face image information, and the like, and these pieces of information can also be used. In addition, the present invention may be used in cooperation with a device such as a wearable terminal used in a state of being in close contact with the user, and data to be noticed, such as arrhythmia, can be easily acquired by the device. The health problem affecting the feet can be detected also in accordance with the pattern of the acceleration sensor during walking. By performing analysis using a history pattern including a plurality of data, instead of analysis of single data that may include errors depending on the device or physical condition, diet, or the condition of a living scene, it is possible to provide information such as the presence or absence or possibility of a disease, recovery, a time to visit a hospital, and advice information with high accuracy. If the accuracy of the information is low, the user is delayed in receiving the visit, and unnecessary fear arises.

In many proposals so far, the measures against such accuracy are insufficient, but in the present embodiment, information that the subject can reasonably go to a medical institution or the like can be provided in consideration of the accuracy and the situation of the subject. Since the information of the facility that can receive the examination or treatment for grasping the accurate health state can be provided, the user can accept the treatment or improve the lifestyle habits by the health grasping, and a healthier life is passed.

In one embodiment of the present invention, the control unit 1 is described as an IT device including a CPU, a memory, an HDD, and the like. However, in addition to being configured in the form of software by a CPU and a program, a part or all of each unit may be configured by a hardware circuit, a hardware configuration such as a gate circuit generated based on a program language described by a hardware description language (Verilog), or a hardware configuration using software such as a DSP (Digital Signal Processor) may be used. They may of course also be combined appropriately.

The control unit 1 is not limited to a CPU, and may be any element as long as it realizes a function as a controller, and the processing of each unit described above may be performed by 1 or more processors configured as hardware. For example, each unit may be a processor configured as an electronic circuit, or may be a circuit unit in a processor configured as an integrated circuit such as an FPGA (Field Programmable Gate Array). Alternatively, a processor including 1 or more CPUs may read and execute a computer program recorded in a recording medium to execute the functions of each unit.

In addition, in the technique described in the present specification, the control mainly described in the flowchart may be set in the form of a program in many cases, and may be recorded in a recording medium or a recording unit. The recording method to the recording medium or the recording unit may be performed at the time of shipment of the product, may be performed using a distributed recording medium, or may be downloaded via the internet.

In addition, although the operation in the present embodiment is described using the flowchart in one embodiment of the present invention, the order of the processing steps may be changed, arbitrary steps may be omitted, steps may be added, and the specific processing content in each step may be changed.

In the operation flows in the claims, the description, and the drawings, even if the description is made using a language expressing the order of "first", "next", and the like for convenience, it is not intended that the description be carried out in this order in a portion not specifically described.

The present invention is not limited to the above-described embodiments, and the components can be modified and embodied in the implementation stage without departing from the gist thereof. Further, various inventions can be formed by appropriate combinations of a plurality of constituent elements disclosed in the above embodiments. For example, some of all the components shown in the embodiments may be deleted. Further, the constituent elements in the different embodiments may be appropriately combined.

Description of the reference numerals

1: control unit, 1 a: communication control unit, 1 b: ID determination unit, 1 c: information providing unit, 1 d: inference model specification determination unit, 1 e: inference delegation unit, 1 f: search unit, 2 a: 1 st device, 2 b: device 2, 4: a terminal, 5: learning unit, 5 a: input/output modeling unit, 5 b: specification comparison section, 6: learning delegation unit, 6 a: recording unit, 6 b: object type a image group, 6 c: training data, 6 d: specification setting unit, 6 e: communication unit, 6 f: control unit, 7: inference engine, 8: DB section, 8 a: a list of the history of different IDs.

Claims (9)

1. An information transmission device, characterized in that,

the information transmission device includes:

a 1 st inspection data acquisition unit that acquires a 1 st inspection data set of a subject person obtained by the 1 st device as a time series; and

a 2 nd inspection data acquisition unit configured to acquire a 2 nd inspection data set of the subject person as a time series, the 2 nd inspection data set being acquired by a 2 nd device, the 2 nd device being capable of performing an inspection capable of complementing the 1 st inspection data set; and

a delivery information determining unit that determines delivery information to be provided to the subject using the 1 st inspection data group and the 2 nd inspection data group,

the 1 st inspection data set and the 2 nd inspection data set complement each other for inspection time or inspection items.

2. The information delivery apparatus according to claim 1,

the delivery information determining unit determines the delivery information according to an inference model obtained by learning according to a change pattern of an inspection data set acquired by a plurality of devices.

3. The information delivery apparatus according to claim 1,

the delivery information determining unit corrects the 1 st inspection data group and the 2 nd inspection data group in accordance with the 1 st inspection data group acquired by the 1 st device and the 2 nd inspection data group acquired by the 2 nd device, calculates the reliability when inference is performed using the corrected inspection data group as input, and determines the delivery information in accordance with the reliability.

4. The information delivery apparatus according to claim 3,

the transmission information determining unit performs a four-way operation on a numerical value common to the respective data included in the inspection data group when performing the correction for each of the inspection data groups.

5. The information delivery apparatus according to claim 1,

the delivery information determining unit corrects the 1 st inspection data group and the 2 nd inspection data group in accordance with the 1 st inspection data group acquired by the 1 st device and the 2 nd inspection data group acquired by the 2 nd device, combines the corrected inspection data groups into 1 inspection data group, inputs the combined inspection data group to an inference model, performs inference, and determines the delivery information based on an inference result.

6. The information delivery apparatus according to claim 1,

the delivery information determining unit corrects the 1 st inspection data group and the 2 nd inspection data group in accordance with the 1 st inspection data group acquired by the 1 st device and the 2 nd inspection data group acquired by the 2 nd device, inputs the corrected inspection data groups to an inference model, and determines the delivery information by comprehensively determining an inference result based on each inference model.

7. The information delivery apparatus according to claim 1,

the 1 st inspection data acquisition unit and the 2 nd inspection data acquisition unit determine whether the inspection data set is from the subject person or from a person other than the subject person, and acquire the inspection data set as the 1 st inspection data set or the 2 nd inspection data set when the inspection data set is from the subject person.

8. The information delivery apparatus according to claim 1,

the 1 st and 2 nd inspection data sets are data obtained from the output result of any one of a color sensor, a shape sensor, a hardness sensor, an olfactory sensor, a gas component sensor, a color change detection sensor when a specific reagent is added, and a shape determination based on an enlarged observation image for defecation.

9. An information delivery method, characterized in that,

the 1 st examination data group of the subject person obtained by the 1 st equipment is acquired as a time series,

acquiring a 2 nd inspection data set of the subject person as a time series obtained by a 2 nd device, the 2 nd device being capable of performing an inspection capable of complementing the 1 st inspection data set,

determining delivery information to be provided to the subject person using the 1 st examination data group and the 2 nd examination data group,

the 1 st inspection data set and the 2 nd inspection data set complement each other for inspection time or inspection items.

Images