JP2018161462A - Information processing method, information processing device, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To accurately grasp a biological state of a measurement object person.SOLUTION: An information processing device acquires measurement data for calculating biological information on a measurement object person from a measuring part that performs measurement in a non-contact manner; classifies the measurement data into groups according to at least a posture of the measurement object person based on the content of the acquired measurement data; calculates the biological information from the classified measurement data; compares the calculated biological information with a reference value corresponding to a group that the measurement data belongs to; and causes an information presentation part to carry out notification on the basis of the comparison.SELECTED DRAWING: Figure 3

Description

  The present disclosure relates to a technique for grasping a biological state of a measurement subject.

  In recent years, technology development for sensing biological information of a measurement subject in a non-contact manner has been performed. For example, in Patent Document 1, for the purpose of performing comfortable air conditioning for a room occupant, an indoor captured image and an indoor temperature distribution image are acquired, the face position is calculated based on the captured image, and the face position is calculated. An air conditioner that acquires temperature information at a position away from a predetermined distance from a temperature distribution image and performs air-conditioning control according to the acquired temperature information is disclosed.

  Patent Document 2 discloses the following technique for the purpose of stably calculating the temperature of the occupant's face in consideration of the orientation of the occupant's face. That is, in Patent Document 2, a passenger's face area is extracted from a temperature distribution map obtained by an IR (InfraRed) camera, and a peripheral area located on the left and right sides of the face area with the central area and the central area in between. A technique is disclosed in which the contribution of each region is adjusted such that the area of the occupant is equal, and the face temperature of the occupant is calculated based on the adjusted contribution.

Japanese Unexamined Patent Publication No. 2009-11989 Japanese Patent Laying-Open No. 2015-55393

  However, in the prior art, it may be difficult to accurately grasp the biological state of the measurement subject. For example, it is known that the temperature of the body part on the paralyzed side is extremely low in a human body with half body paralysis. Therefore, as disclosed in Patent Document 2, just by measuring the body temperature of the face by adjusting the contribution of each region so that the areas of the central region and the peripheral region are equal, the change in the body temperature of the human body can be reduced. It is difficult to grasp accurately.

  Since the invention disclosed in Patent Document 1 detects the temperature at a position away from the face position by a predetermined distance, it is difficult to accurately grasp the body temperature of the human body itself.

  The objective of this indication is providing the technique which can grasp | ascertain the biological condition of a measurement subject correctly.

An information processing method according to an aspect of the present disclosure uses a computer to acquire measurement data for calculating biological information of a measurement target person from a measurement unit that performs measurement without contact,
Based on the content of the acquired measurement data, the measurement data is classified into a group corresponding to at least the posture of the measurement subject, the biological information is calculated from the classified measurement data, and the calculated biological information To the reference value corresponding to the group to which the measurement data belongs,
A notification based on the comparison is made to be made to an information presenting unit.

  According to the embodiment of the present disclosure, it is possible to accurately grasp the biological state of the measurement subject.

It is a figure which shows the outline | summary of the biometric information sensing apparatus to which the information processing apparatus concerning Embodiment 1 was applied. It is a figure which shows an example of the connection structure of a biometric information sensing apparatus. It is a figure which shows the structure of the function of the biometric information sensing apparatus to which the information processing apparatus concerning Embodiment 1 was applied. It is a figure explaining the data processing method in a learning phase. It is a figure explaining the data processing method in a detection phase. FIG. 6 is a diagram illustrating a data processing method in a detection phase following FIG. 5. It is a flowchart which shows an example of a process of the information processing apparatus in a detection phase. It is a figure which shows the outline | summary of the biometric information sensing apparatus with which the information processing apparatus concerning Embodiment 2 was applied. It is a figure which shows an example of a function structure of the biometric information sensing apparatus with which the information processing apparatus concerning Embodiment 2 was applied. It is a figure explaining the data processing method in a learning phase. It is a figure explaining the data processing method in a detection phase. FIG. 12 is a diagram illustrating a data processing method in a detection phase following FIG. 11. It is a figure which shows the outline | summary of the biometric information sensing apparatus with which the information processing apparatus concerning the modification 2 of this indication was applied. It is a figure which shows the structure of the function of the biometric information sensing apparatus with which the information processing apparatus concerning the modification 3 of this indication was applied. It is a figure showing an example of connection composition of a living body information sensing device to which an information processor concerning modification 3 of this indication was applied. It is a figure which shows the structure of the function of the biometric information sensing apparatus with which the information processing apparatus concerning the modification 4 of this indication was applied. It is a figure which shows the structure of the function of the biometric information sensing apparatus with which the information processing apparatus concerning the modification 5 of this indication was applied. It is a figure which shows the 1st example of the connection structure of the biometric information sensing apparatus to which the information processing apparatus concerning the modification 5 of this indication was applied. It is a figure which shows the 2nd example of the connection structure of the biometric information sensing apparatus with which the information processing apparatus concerning the modification 5 of this indication was applied. It is a figure which shows the connection structure of the biometric information sensing apparatus to which the information processing apparatus concerning the modification 6 of this indication was applied. It is a figure which shows the connection structure of the biometric information sensing apparatus to which the information processing apparatus concerning the modification 7 of this indication was applied. It is a figure which shows the connection structure of the biometric information sensing apparatus to which the information processing apparatus concerning the modification 8 of this indication was applied. It is a figure which shows the outline | summary of the biometric information sensing apparatus with which the information processing apparatus concerning the modification 9 of this indication was applied.

(Knowledge that forms the basis of this disclosure)
In elderly care, it is essential to record and manage changes in living conditions and body temperature when performing daily physical condition management of the elderly. Therefore, a physical condition management system that periodically measures biological information such as the body temperature of a human body, compares the measured value with a reference value, and notifies if there is a change in abnormality or the like has been studied.

  Conventionally, a thermometer and a radiant heat sensor are known as means for measuring the body temperature of a human body. These means grasp the body temperature by measuring specific parts of the human body such as the armpit, the inner ear, and the face front.

  However, when these contact-type means are applied to a physical condition management system, the body temperature is measured by bringing a thermometer into contact with the lower part of the heel, or the body temperature is measured using a dedicated device that makes contact with the human body such as an inner ear thermometer. There is a need to. For this reason, it is necessary to have the subject of measurement stand still, or to measure the body temperature of the subject of measurement with the attendant. As a result, there is a problem that both the measurer and the measurement subject are burdened.

  Therefore, it is conceivable to apply the technique of Patent Document 2 in which body temperature is measured using a thermal image sensor that measures body temperature in a non-contact manner to a system for managing physical condition. Here, in Patent Document 2, since one body temperature is finally calculated by adjusting the contribution of each region, the physical condition management system to which the method of Patent Document 2 is applied detects the presence or absence of a change. You only need to prepare one reference body temperature to do.

  However, the body temperature on the paralyzed side is greatly different from the body temperature on the non-paralyzed side. Therefore, when the presence or absence of a change in body temperature is detected using a single reference body temperature, if the body temperature on the paralyzed side is measured according to the posture that the measurement subject is at the time of measurement, the measurement is performed despite the fact that the body temperature is originally fever For example, since the body temperature is lower than the reference body temperature, it may be erroneously determined as normal heat. In addition, when the body temperature on the side that is not paralyzed is measured by the posture that the measurement subject takes at the time of detection, because the measured body temperature is higher than the reference body temperature, for example, the body temperature is normal, but heat is generated. May be erroneously determined.

  This indication is made in order to solve such a problem, and provides the technique which can grasp | ascertain the biological condition of a measuring object person correctly, without putting a burden on both a measuring object person and a measurement person. That is.

An information processing apparatus according to an aspect of the present disclosure uses a computer to acquire measurement data for calculating biological information of a measurement target person from a measurement unit that performs measurement without contact,
Based on the content of the acquired measurement data, the measurement data is classified into a group corresponding to at least the posture of the measurement subject, the biological information is calculated from the classified measurement data, and the calculated biological information To the reference value corresponding to the group to which the measurement data belongs,
A notification based on the comparison is made to be made to an information presenting unit.

  According to this aspect, the measurement data measured by the measurement unit is classified into groups based on the posture of the measurement target person, the reference value corresponding to the classified group is compared with the classified measurement data, and compared. Based notifications are made. Therefore, the notification can be made after accurately grasping the biological state of the measurement subject regardless of the posture of the measurement subject taken at the time of measurement. In other words, it is possible to reduce the occurrence of processing for erroneous notification. Therefore, it is possible to reduce the processing load for notification. Moreover, according to this aspect, since the biological information is measured in a non-contact manner, it is possible to grasp the biological state without imposing a burden on both the measurement subject and the measurement person.

In the above aspect, in the classification of the measurement data, the measurement data is classified into the group using a machine learning model in which the measurement data is classified into the group based on the content of the measurement data.
The reference value may be a representative value of the biological information of the group calculated from the measurement data classified into the group in the machine learning.
Thereby, since the measurement data is classified by the machine learning model, more appropriate classification can be performed as compared with the rule-based classification. Further, since the reference value is determined from the actually classified measurement data, a more accurate reference value can be used for the comparison process.

In the above aspect, the information processing method further includes:
Classifying the measurement data into the group based on the content of the measurement data in the machine learning model,
In the machine learning, a representative value of the biological information of the group may be calculated from the measurement data classified into the group as the reference value of the group.

  According to this aspect, since the representative value of the biological information for each group is calculated by machine learning from the measurement data classified for each group, and the representative value is calculated as the reference value, the group suitable for the measurement target person Each reference value can be calculated.

In the above aspect, the measurement unit is a thermal image sensor,
The biological information is body temperature,
The measurement data is thermal image data obtained by the thermal image sensor,
In the classification of the measurement data, the thermal image data is classified into the group based on the characteristics of the region representing the measurement subject included in the thermal image data obtained by the thermal image sensor,
In the comparison, the body temperature of the measurement subject is calculated from the classified thermal image data, and the calculated body temperature is compared with a reference value corresponding to a group to which the classified thermal image data belongs. Good.

  According to this aspect, since the measurement part is comprised with the thermal image sensor, it can grasp | ascertain the biological condition of a measurement subject correctly.

In the above aspect, the measurement unit is a radio wave sensor,
The biological information is an activity amount including at least one of a body motion value, a respiratory rate, and a heart rate of the measurement subject,
The measurement data is activity amount data indicating the activity amount obtained by the radio wave sensor,
In the classification of the measurement data, the activity data is classified into the group based on waveform information of the activity data obtained by the radio wave sensor,
In the comparison, the activity amount of the measurement subject is calculated from the classified activity amount data, and the calculated activity amount is compared with a reference value corresponding to a group to which the classified activity amount data belongs. May be.

  According to this aspect, since the measurement unit is configured by the radio wave sensor, it is possible to accurately detect at least one of the body motion value, the respiratory rate, and the heart rate of the measurement subject.

  In the above aspect, when the number of the measurement data that does not correspond to any of the groups exceeds a reference number, information that prompts relearning of the machine learning model may be generated.

  According to this aspect, if the number of measurement data that does not correspond to any posture classified by machine learning exceeds the reference number during execution of the process for comparing the measurement data with the reference value, the machine learning model Information that encourages relearning is generated. Therefore, for example, an administrator who has recognized this information can re-execute machine learning. As a result, even if the measurement subject comes to take a posture that was not seen during machine learning during the process of comparing the measurement data with the reference value, the biological state of the measurement subject can be accurately grasped.

In the above aspect, the information processing method further acquires time data indicating a measurement time for the measurement data,
In the comparison, it is determined whether or not the measurement data is measured in a predetermined time zone based on the time data. When it is determined that the measurement data is measured in the predetermined time zone, the comparison is performed. May be executed.

  According to this aspect, when the measurement data belongs to a predetermined time zone, the comparison process is executed. For example, the measurement data measured during the time zone in which the state of the measurement subject is stable is used. Can be compared, and biological information can be grasped more accurately.

In the above aspect, the posture may be a body orientation of the measurement subject.
Thereby, measurement data can be classified into groups according to the direction of the body of the measurement subject. Here, since the characteristics of the body such as blood flow vary depending on the orientation of the body, the measurement data, that is, the reference value of biological information such as body temperature also varies depending on the orientation of the body. Therefore, it can be determined whether there is an abnormality in the measurement subject using an appropriate reference value.

In the above aspect, the notification based on the comparison may be a comparison result between the biological information and the reference value, or an abnormality of the measurement subject.
Thereby, the person who can receive the notification of the information presenting unit can be made aware of the abnormality of the measurement subject.

  The said aspect is realizable as a method which uses the process means to comprise as a step. And this indication is realizable as a program which makes a computer perform the step contained in the method. Furthermore, this indication is realizable as computer-readable storage media, such as CD-ROM which memorize | stored the program.

(Embodiment 1)
FIG. 1 is a diagram illustrating an outline of a biological information sensing device to which the information processing apparatus according to the first embodiment is applied. In FIG. 1, a sensing device 101 includes a thermal image sensor and is disposed in a living room such as a bedroom in which a bedding 107 is disposed. The sensing device 101 is installed side by side with the air conditioner 105, for example. It is preferable to arrange the sensing device 101 in an inconspicuous position, for example, in the vicinity of the side surface of the air conditioner 105. The sensing device 101 is installed in a living room, for example, so that the entire area of the bedding 107 is included in the measurement range.

  The sensing device 101 acquires thermal image data indicating a heat distribution in a room including a human body 106 (an example of a measurement target person). The sensing device 101 transmits the acquired thermal image data to the information processing device 102 via a gateway (hereinafter referred to as GW) 104. The information processing apparatus 102 manages thermal image data by classifying it into groups based on the posture of the human body 106. The GW 104 is also referred to as a router.

  Here, the living room may be, for example, one room where the human body 106 who is a care recipient stays in an elderly care facility, or may be one room where the human body 106 lives.

  When the information processing apparatus 102 detects a change in the body temperature of the human body 106 by comparing the thermal image data classified for each group with the thermal image data measured this time, the body temperature of the human body 106 is abnormally detected via the GW 104. Is transmitted to the information display apparatus 103. When the information display device 103 receives the alerting information, the information display device 103 issues an alert. Thereby, the administrator of the human body 106 recognizes that the body temperature of the human body 106 is abnormal.

  As a manager of the human body 106, for example, a caregiver who cares for the human body 106 or a manager of a care facility can be employed.

  FIG. 2 is a diagram illustrating an example of a connection configuration of the biological information sensing device. As shown in FIG. 2, the biological information sensing device includes a sensing device 101, an information processing device 102, an information display device 103, and a GW 104. The sensing device 101, the information processing device 102, and the information display device 103 are each connected to the GW 104 via a predetermined network. As the predetermined network, for example, a wired LAN, a wireless LAN, or a LAN in which both are mixed is adopted.

  FIG. 3 is a diagram illustrating a functional configuration of the biological information sensing device to which the information processing apparatus according to the first embodiment is applied. The biological information sensing device includes a sensing unit 101A corresponding to the sensing device 101, an information processing unit 102A corresponding to the information processing device 102, and an information display unit 103A corresponding to the information display device 103.

  The sensing unit 101A (an example of a measurement unit) includes a sensor unit 301 and a transmission unit 302, and measures biological information without contact. The sensor unit 301 is composed of, for example, a thermal image sensor, and acquires thermal image data (an example of measurement data) by imaging a room of the human body 106 at a predetermined sampling period. The transmission unit 302 is configured by a wireless LAN or wired LAN communication circuit, for example, and transmits the thermal image data acquired by the sensor unit 301 to the information processing unit 102A at a predetermined sampling period.

  The information processing unit 102 </ b> A includes a reception unit 303, a time data adding unit 304, a data classification unit 305, a data management unit 306, a change detection unit 307, a transmission unit 308, and a DB unit 311.

  The reception unit 303 (an example of an acquisition unit) is configured by, for example, a wireless LAN or wired LAN communication circuit, and receives thermal image data transmitted from the sensing unit 101A at a predetermined sampling period.

  The time data assigning unit 304 assigns a measurement time to the thermal image data received by the receiving unit 303. Here, as the measurement time, for example, the acquisition time when the reception unit 303 acquires the thermal image data is given. The measurement time is constituted by data representing year / month / day / hour / minute / second, for example.

  The data classification unit 305 acquires a plurality of thermal image data by sequentially acquiring the thermal image data to which the time data is added by the time data adding unit 304 in the learning phase for learning the group based on the posture. Then, the data classification unit 305 learns a group by classifying the thermal image data based on the characteristics of the region representing the human body 106 included in each of the plurality of acquired thermal image data. As a feature of the pixel, an outline of a region representing the human body 106 and a temperature distribution constituting the region are employed.

  Here, the data classification unit 305 may learn a group by machine learning, for example. Here, as machine learning, for example, supervised machine learning or unsupervised machine learning using a neural network may be employed. In the case of supervised machine learning, for example, the data classifying unit 305 learns a weighting factor of a neural network so that a group identifier for identifying a group based on a posture given in advance to each thermal image data is output. That's fine.

  Here, for example, the orientation of the human body 106 can be adopted as the posture given in advance. As the orientation of the human body 106, for example, “front” indicating the orientation of the human body 106 as viewed from the front, “left” indicating the orientation of the human body 106 as viewed from the left with respect to the front, and the human body 106 with respect to the front. At least four patterns of “right” indicating the direction viewed from the right side and “rear” indicating the direction of the human body 106 viewed from the back can be adopted. Furthermore, as the orientation of the human body 106, for example, a depth pattern in each of the four patterns of front, rear, left, and right may be added. As the depth pattern, for example, a pattern in which the distance from the thermal image sensor is divided in stages can be adopted. In this case, as the depth pattern, for example, a pattern such as “near”, “normal”, and “far” can be adopted. Assuming that the human body 106 has four patterns of front, back, left, and right, and the human body 106 has three patterns of “near”, “normal”, and “far”, the thermal image data is 4 × depending on the posture and depth. It will be classified into 12 groups of 3 = 12. Here, the depth is taken into consideration because the measurement accuracy of the thermal image sensor decreases as the distance from the thermal image sensor to the human body 106 increases. As the orientation of the human body 106, a direction other than the above four patterns (for example, right diagonally forward, left diagonally forward, right diagonally backward, and left diagonally rear) may be added, or any one of the above four patterns One or more may be omitted.

  When employing unsupervised machine learning, the data classifying unit 305 may employ a deep neural network having a higher number of layers than a normal neural network. When the deep neural network is adopted, the data classification unit 305 classifies the thermal image data according to the characteristics of the region indicating the human body 106 included in the thermal image data without setting the posture and depth in advance. Can learn groups.

  Note that the data classification unit 305 may learn a group using a method other than machine learning. When employing a method other than machine learning, the data classification unit 305 first extracts the region of the human body 106 from the thermal image data. In this case, for example, the data classification unit 305 calculates a HOG (Histograms of Oriented Gradients) feature amount from the thermal image data, and divides the thermal image data into a region of the human body 106 and a background region according to the HOG feature amount. The region of the human body 106 may be extracted. Next, the data classification unit 305 may determine the orientation of the human body 106 using the extracted outline of the region of the human body 106 and the relative positional relationship between the nose and the mouth in the face. Here, the face can be extracted from the shape of the human body 106, and the positions of the nose and mouth can be obtained from the heat distribution of the face. Here, the determined orientation of the human body 106 is a predetermined orientation such as the above-described four patterns of front, rear, left, and right.

  Next, the data classification unit 305 classifies the thermal image data by assigning a group identifier corresponding to the determined orientation to the thermal image data. In this case, the data classification unit 305 may further classify the thermal image data for each of the depth patterns based on the size of the face. In this case, the data classification unit 305 may classify the thermal image data by adding a group identifier for each set of the orientation and depth of the human body 106 to the thermal image data.

  On the other hand, the data classification unit 305 sequentially acquires the thermal image data to which the time data is added by the time data adding unit 304 in the detection phase in which a change in the biological information is detected. Then, the data classification unit 305 classifies the acquired thermal image data into any group learned in the learning phase based on the characteristics of the region representing the human body 106 included in each of the acquired thermal image data. For example, when a neural network is employed in the learning phase, the data classification unit 305 determines the group to which the thermal image data belongs by inputting the thermal image data to be detected into the neural network in the detection phase. Good.

  Further, when a method other than the above-described machine learning is adopted in the learning phase, the data classification unit 305 may determine a group to which the detection target thermal image data belongs using the method.

  In the learning phase, the data management unit 306 calculates a representative value of body temperature for each group from the thermal image data classified by the data classification unit 305 as a reference value for each group. Here, the reference value assumes normal heat of the human body 106, and an average value of body temperature for each group obtained from the classified thermal image data is adopted.

  The data management unit 306 manages the reference value by storing the calculated reference value for each group in the DB unit 311. Note that the data management unit 306 may update the reference value based on the thermal image data classification result even in the detection phase.

  The change detection unit 307 (an example of a comparison unit) calculates the body temperature of the human body 106 (an example of biological information) from the thermal image data classified into one of the groups by the data classification unit 305 in the detection phase, and calculates the calculated body temperature. Is compared with a reference value corresponding to the group to which the thermal image data belongs, to detect the presence or absence of a change in body temperature of the human body 106. And the change detection part 307 produces | generates the alerting information which shows abnormality of body temperature, when the change of body temperature is detected. Here, since the reference value assumes normal heat of the human body 106, the change detection unit 307 has a temperature difference between the body temperature calculated from the target thermal image data and normal heat indicated by the reference value, for example, plus or minus 1. What is necessary is just to determine that there was a change in body temperature if it is equal to or higher. Here, if the temperature difference is plus or minus 1 ° C. or more, it is determined that there is a change in body temperature, but the present disclosure is not limited to this, and a predetermined temperature other than 1 ° C. (for example, 0. If it is 5 ° C., 1.5 ° C. or more), it may be determined that there is a change in body temperature.

  Here, the change detection unit 307 may detect a part (for example, a face) specified by a comparison part parameter 311C described later from the thermal image data, and calculate the body temperature of the human body 106, for example, the average temperature of the detected part. .

  The transmission unit 308 is configured by, for example, a wireless LAN or wired LAN communication circuit. When the alerting information is generated by the change detection unit 307, the transmission unit 308 transmits the alerting information to the information display unit 103A.

  The database (hereinafter referred to as “DB”) unit 311 includes, for example, a non-volatile storage device, and records various data managed by the data management unit 306. The DB unit 311 stores classification parameters 311A, classification data 311B, and comparison site parameters 311C.

  For example, if a neural network is employed as the learning phase, the classification parameter 311A is a weighting coefficient of the neural network obtained as a result of learning.

  The classification data 311B is thermal image data classified for each group by the data classification unit 305 in the learning phase and the detection phase.

  The comparison part parameter 311C is a parameter indicating the part of the human body 106 to be compared that is used by the change detection unit 307 when detecting the change. For example, if a face is a comparison target, information indicating the face is adopted as the comparison site parameter.

  Note that when supervised machine learning is employed, the DB unit 311 may store information indicating postures and depths to be classified in advance.

  For example, the information display unit 103 </ b> A is configured by a computer that can be used by the administrator of the human body 106, and includes a receiving unit 309 and a display unit 310, and outputs alerting information. Here, 103 A of information display parts may be comprised with a stationary computer, and may be comprised with portable computers, such as a tablet terminal and a smart phone.

  The receiving unit 309 includes a wireless LAN or wired LAN communication circuit, and receives alerting information transmitted from the information processing unit 102A. The display unit 310 is composed of, for example, a liquid crystal display or an organic EL display, and displays alerting information. In this case, the display unit 310 may display, for example, a message indicating that the body temperature of the human body 106 is abnormal. Note that the information display unit 103A may output a sound indicating that the body temperature of the human body 106 is abnormal from a speaker (not shown).

  Next, a data processing method in the information processing unit 102A will be described with reference to FIGS. FIG. 4 is a diagram for explaining a data processing method in the learning phase. FIG. 5 is a diagram for explaining a data processing method in the detection phase. FIG. 6 is a diagram for explaining the data processing method in the detection phase following FIG.

  Please refer to FIG. When the biological information sensing device is installed at the site, the information processing unit 102A generates the classification parameter 311A by learning the thermal image data for a certain period. First, the time data adding unit 304 adds time data to each of the thermal image data (401) acquired by the sensing unit 101A (402). Next, the data classification unit 305 performs a mask process for masking a region other than a predetermined temperature range (for example, 25 ° C. to 45 ° C.) for each of the thermal image data (403) to which the time data is given. Next, the data classification unit 305 classifies the thermal image data for each group of the human body 106 by performing machine learning using a neural network, and generates a classification parameter 311A (404). Here, 25 ° C. to 45 ° C. is adopted as the predetermined temperature range. However, by adopting a temperature range of 30 ° C. to 45 ° C., for example, the human body 106 and an object other than the human body 106 (for example, furniture) The difference can be clarified more. The predetermined temperature range assumes a temperature range that the human body 106 can take.

  Next, referring to FIG. When the learning phase ends, the detection phase starts. First, the time data assigning unit 304 assigns time data to the thermal image data (501) acquired by the sensing unit 101A as in the learning phase (402). Next, as in the learning phase, the data classifying unit 305 performs a mask process on the thermal image data (503) to which time data is added, and masks a region other than the predetermined temperature range. Next, the data classification unit 305 inputs any one of the thermal image data classified in the learning phase by inputting the thermal image data in which the region other than the predetermined temperature range is masked to the neural network generated in the learning phase. (504). Here, since there are the first to Nth groups as the groups, the thermal image data is classified into any of the first to Nth groups (N is an integer of 1 or more). Thereby, the thermal image data to be detected is classified based on the characteristics of the region representing the human body 106. The classified thermal image data is stored in the DB unit 311 as classification data 311B. Note that the neural network generated by the learning phase is an example of a machine learning model.

  Next, refer to FIG. The change detection unit 307 reads out thermal image data to be detected from the classification data 311B, and detects a site specified by the comparison site parameter 311C from the read thermal image data (601). For example, if a face is designated as the comparison site parameter 311C, the change detection unit 307 detects the face from the thermal image data to be detected.

  Next, the change detection unit 307 calculates the body temperature of the part detected from the thermal image data, and compares the calculated body temperature with the reference value corresponding to the group to which the thermal image data belongs, thereby calculating the body temperature of the human body 106. The presence or absence of change is detected (602). Here, the change detection unit 307 determines that there has been a change in body temperature if the temperature difference between the calculated body temperature and the normal temperature indicated by the reference value is greater than or equal to ± 1 ° C.

  In the example of FIG. 6, the classification data 311B is divided into first to Nth groups according to posture and depth. Therefore, there are N reference values corresponding to the first to Nth groups.

  In the present embodiment, the reference value calculated in the learning phase is employed, but the present disclosure is not limited to this, and a value determined in advance for each group may be employed. Alternatively, as the reference value, a value calculated for each group from the thermal image data classified in the past certain period from the present may be adopted.

  FIG. 7 is a flowchart illustrating an example of processing of the information processing apparatus 102 in the detection phase. Note that the flowchart of FIG. 7 is periodically executed, for example, at a predetermined sampling period in which the sensing unit 101A acquires thermal image data.

  First, the data classifying unit 305 acquires thermal image data to which time data is added by the time data adding unit 304 (S701). Next, the data classification unit 305 performs mask processing on the acquired thermal image data, and masks portions other than 25 ° C. to 45 ° C. (S702).

  Next, the data classification unit 305 classifies the thermal image data subjected to the mask processing by inputting it into the neural network in which the classification parameter 311A is generated in the learning phase (S703). Accordingly, the thermal image data is classified into one of the first to Nth groups according to the posture and depth. Further, it may occur that the thermal image data cannot be classified into any of the first to Nth groups. For example, this is a case where the human body 106 comes to take the posture and depth not taken in the learning phase in the detection phase. In this case, the thermal image data is classified into other groups.

  Next, the data classification unit 305 determines whether or not the number of thermal image data classified into other groups exceeds the reference number (S704).

  If the number of thermal image data classified into other groups exceeds the reference number (YES in S704), a flag that prompts re-execution of the learning phase is generated (S705). Here, the generated flag is transmitted to, for example, the information display unit 103A. 103 A of information display parts display the image which shows that the flag was produced | generated on the display part 310. FIG. As a result, the manager is prompted to execute the learning phase again. For example, the administrator inputs a relearning command to the information processing unit 102A using an input device (not shown). Then, the data classification unit 305 executes the learning phase using all the thermal image data accumulated as the classification data 311B so far, and generates the classification parameter 311A again. Accordingly, a group including the posture and depth of the human body 106 indicated by the thermal image data classified into other groups is newly learned. As a result, a group based on the posture and depth of the human body 106 newly taken in the detection phase is added as a group to be classified. In addition, when the learning phase is re-executed, since the reference value for each group is calculated from the thermal image data classified for each group, the thermal image data classified into the newly added group is the corresponding standard. The change in temperature is detected by comparison with the value.

  Next, the change detection unit 307 detects the region indicated by the comparison region parameter 311C from the thermal image data to be detected, and calculates the body temperature of the human body 106 using the pixel value of the detected region (S706).

  Next, the change detection unit 307 compares the reference value corresponding to the group of thermal image data to be detected with the body temperature of the human body 106 calculated in S706, and the temperature difference between the body temperature of the human body 106 and the reference value is determined. If it is greater than or equal to plus or minus 1 ° C. (YES in S708), alerting information is generated (S709). When S709 ends, the process proceeds to S710. The generated alerting information is transmitted to the information display unit 103A, and an alert is issued to the administrator. On the other hand, if the temperature difference between the body temperature of human body 106 and the reference value is less than plus or minus 1 ° C. (NO in S708), the process proceeds to S710.

  In S710, the data management unit 306 updates the reference value of the group to which the thermal image data belongs using the classification data 311B in which the thermal image data to be detected is classified, and returns the process to S701. For example, if the thermal image data to be detected is classified into the first group, the data management unit 306 may update the reference value of the first group.

  As described above, according to the information processing apparatus 102 of Embodiment 1, the thermal image data is classified according to the group based on the posture and the depth, and is calculated from the reference value corresponding to the group and the thermal image data. By comparing with the body temperature, an abnormality in the body temperature of the human body 106 is detected. Therefore, the information processing apparatus 102 can accurately detect the presence or absence of an abnormal body temperature of the human body 106 regardless of the posture and depth of the human body 106 taken at the time of measurement. In addition, since the information processing apparatus 102 can measure the body temperature of the human body 106 in a non-contact manner, both the human body 106 and the manager of the human body 106 can detect an abnormality in the body temperature. Further, in this aspect, since the reference value is calculated from the thermal image data obtained by photographing the human body 106, the reference value suitable for the human body 106 can be calculated.

(Embodiment 2)
FIG. 8 is a diagram illustrating an outline of a biological information sensing device to which the information processing apparatus according to the second embodiment is applied. The biological information sensing device according to the second embodiment employs a sensing device 901_A including a radio wave sensor. In the present embodiment, the description of the same components as those in the first embodiment is omitted.

  The sensing device 901 </ b> _A includes a radio wave sensor and is disposed in a living room such as a bedroom as in FIG. 1. Here, the sensing device 901 </ b> _A is installed in the living room so that the entire range of the bedding 107 is included in the measurement range, for example. The sensing device 901 </ b> _A acquires activity amount data of the human body 106. The sensing device 901_A transmits the acquired activity amount data to the information processing device 902 via the GW 104. The information processing apparatus 902 classifies and manages the activity amount data for each group based on the posture of the human body 106.

  When the information processing device 902 detects a change in the activity amount (an example of biological information) of the human body 106 by comparing the activity amount data classified for each group with the activity amount data measured this time, the information processing device 902 transmits the change through the GW 104. The alerting information indicating that the activity amount of the human body 106 is abnormal is notified to the information display device 103. When the information display device 103 receives the alerting information, the information display device 103 issues an alert. Thereby, the manager of the human body 106 recognizes an abnormality in the activity amount of the human body 106.

  FIG. 9 is a diagram illustrating an example of a functional configuration of the biological information sensing device to which the information processing apparatus according to the second embodiment is applied. The biological information sensing device includes a sensing unit 901A (an example of a measurement unit) corresponding to the sensing device 901_A, an information processing unit 902A corresponding to the information processing device 902, and an information display unit 103A corresponding to the information display device 103. The sensing unit 901A is different from the sensor unit 901 in FIG.

  The sensor unit 901 includes, for example, a 2ch Doppler type radio wave sensor. The human body 106 is irradiated with radio waves at a predetermined sampling period and receives a reflected wave from the human body 106, thereby Acquire activity data. As the radio wave, for example, a microwave of 24 GHz band can be adopted.

  The sensor unit 901 may employ a radio wave sensor other than the Doppler method. For example, a radio wave sensor such as a Frequency Modulated Continuous Wave (FMCW) method may be employed.

  In the learning phase, the data classification unit 905 acquires a plurality of activity amount data to which time data is added by the time data addition unit 304, and classifies the acquired plurality of activity amount data on the basis of the waveform information. A group based on 106 postures is learned.

  If the sensor unit 901 is composed of a 2-channel Doppler radio wave sensor, the distance to the human body 106 changes according to the amplitude of the activity data, and the speed of the body movement according to the change in the frequency of the activity data. Changes. Therefore, in this case, amplitude and frequency are employed as the waveform information.

  Further, if the sensor unit 901 is configured by an FMCW radio wave sensor, the distance to the human body 106 changes according to the amplitude and phase of the activity data, and the body changes according to the change in the phase of the activity data. The speed of movement changes. Therefore, in this case, amplitude and phase are adopted as the waveform information.

  In addition, the speed of body movement changes according to the posture of the human body 106 such as lying on the back or lying down, and the speed of body movement changes depending on the state of the human body 106 such as sleeping or waking up. Therefore, by classifying the activity amount data based on the waveform information, the activity amount data can be classified into groups of posture, state, and distance of the human body 106, and groups of posture, state, and distance taken by the human body 106 can be classified. Can learn.

  In the present embodiment, the data classification unit 905 may learn a group using the machine learning described in the first embodiment.

  In the detection phase for detecting a change in biological information, the data classification unit 905 assigns the activity amount data to which the time data is added by the time data adding unit 304 to any group learned in the learning phase based on the waveform information. Classify. Specifically, the data classification unit 905 may classify the activity amount data by inputting the activity amount data to the neural network generated in the learning phase.

  In the learning phase, the data management unit 906 calculates a representative value of the activity amount for each group from the activity amount data classified for each group by the data classification unit 905 as a reference value for each group. Here, as the reference value, the reference values of the body motion value, the heart rate, and the respiration rate can be adopted. In the activity amount data, the frequency band of the body motion value, the heart rate, and the respiration rate is known in advance, so that the body motion value, the heart rate, and the respiration rate can be detected from the values of the respective frequency bands. Therefore, the data management unit 906 may calculate the average value for each group of the body motion value, the heart rate, and the respiration rate as the reference value for the body motion value, the heart rate, and the respiration rate for each group. In the detection phase, the data management unit 306 may update the reference value based on the classification result of the activity amount data.

  In the detection phase, the change detection unit 907 calculates the activity amount including the body motion value, the respiration rate, and the heart rate of the human body 106 from the activity amount data classified into one of the groups by the data classification unit 905. By comparing the amount of activity with the reference value of the corresponding group, the presence or absence of a change in the amount of activity of the human body 106 is detected. And the change detection part 907 produces | generates the alerting information which shows that an active mass is abnormal, when the change of an active mass is detected. Here, since the reference value assumes a normal activity amount of the human body 106, the change detection unit 907 has a difference between the activity amount indicated by the target activity amount data and the reference value greater than or equal to a predetermined value. If so, it may be determined that there is a change in the amount of activity.

  The DB unit 911 stores classification parameters 911A, classification data 911B, and comparison value calculation parameters 911C. The classification parameter 911A is, for example, a weighting factor of a neural network obtained by learning activity amount data in the learning phase. The comparison value calculation parameter 911C is a parameter that determines the content of the comparison value detected from the activity amount data when the change detection unit 907 detects a change. In this embodiment, since the body motion value, the pulse rate, and the heart rate are adopted as the comparison values, the comparison value calculation parameter 911C includes, for example, information indicating each of the body motion value, the pulse rate, and the heart rate. Adopted. Alternatively, as the comparison value calculation parameter 911C, for example, frequency bands in the activity amount data of the body motion value, the pulse rate, and the heart rate may be employed.

  In addition, when supervised machine learning is adopted, the DB unit 911 includes information regarding groups to be classified in advance (information indicating posture (such as lying on the back and leaning), state (sleeping and waking up, etc.), (Distance) may be stored.

  Next, a data processing method in the information processing unit 902A will be described with reference to FIGS. FIG. 10 is a diagram for explaining a data processing method in the learning phase. FIG. 11 is a diagram for explaining a data processing method in the detection phase. FIG. 12 is a diagram for explaining the data processing method in the detection phase following FIG.

  Please refer to FIG. When the biological information sensing device is installed at the site, the information processing unit 902A generates the classification parameter 911A by learning the activity amount data. First, the time data assigning unit 304 assigns time data to each of the activity amount data (1001) acquired by the sensing unit 901A (1002).

  Next, the data classification unit 905 classifies the activity data (1003) to which time data is assigned by group by using machine learning using a neural network, and generates a classification parameter 911A (1004).

  Next, referring to FIG. When the learning phase ends, the detection phase starts. Similar to the learning phase, the time data adding unit 304 adds time data to the activity amount data (1101) acquired by the sensing unit 901A (1002). Next, as in the learning phase, the data classification unit 905 inputs the activity data (1103) to which the time data is assigned to the neural network generated in the learning phase, so that the activity data is classified in the learning phase. The group is classified into one of the groups (1104). Thereby, the activity amount data to be detected is classified based on the waveform information. The classified activity data is stored in the DB unit 911 as classification data 911B.

  Next, refer to FIG. The change detection unit 907 reads the activity amount data to be detected from the classification data 911B, and detects the comparison value specified by the comparison value calculation parameter 911C from the read activity amount data (1201). Here, a body motion value, a pulse rate, and a heart rate are employed as the comparison value. Therefore, the body motion value, the pulse rate, and the heart rate are extracted from the read activity amount data.

  Next, the change detection unit 907 compares the body motion value, the pulse rate, and the heart rate detected from the activity amount data with the reference value corresponding to the group into which the activity amount data is classified, thereby comparing the human body 106. The presence or absence of a change in the amount of activity is detected (1202). Here, the change detection unit 307 determines that the amount of activity has changed if the detected body motion value, the pulse rate, and the difference between the heart rate and the corresponding reference value are greater than or equal to a predetermined value. .

  In the example of FIG. 12, the classification data 911B is divided into first to Nth (N is an integer of 1 or more) N groups. Therefore, there are N reference values corresponding to the first to Nth groups. In the example of FIG. 12, any one of the body motion value, the heart rate, and the pulse rate is illustrated as the reference value. However, in actuality, each of the body motion value, the heart rate, and the pulse rate is illustrated. The reference value is included.

  In the present embodiment, the reference value calculated in the learning phase is employed, but the present disclosure is not limited to this, and a value determined in advance for each group may be employed. Alternatively, as the reference value, a value calculated from activity amount data classified in a certain period in the past from the present may be adopted.

  In the second embodiment, the flowchart is the same as that shown in FIG. 7 except that thermal image data is used as activity amount data, and a description thereof will be omitted.

  As described above, in the second embodiment, the activity amount data is classified for each group including posture, state, and distance by the waveform information of the activity amount data, and the activity amount of the classified activity amount data corresponds to the group. An abnormal amount of activity is detected by comparison with the reference value. Therefore, in the second embodiment, it is possible to accurately detect the presence or absence of an abnormality in the activity amount of the human body 106 regardless of the posture of the human body 106 taken at the time of measurement. Moreover, since this aspect measures biometric information in a non-contact manner, it is possible to detect an abnormality in the amount of activity without imposing a burden on the human body 106 and the measurer.

  The information processing apparatus according to the present disclosure can employ the following modifications.

(Modification 1)
The change detection unit 307 determines whether the measurement data is measurement data measured in a predetermined time zone based on the time data given to the measurement data, and uses the measurement data measured in the predetermined time zone. In some cases, a process for detecting a change may be performed. Here, as the predetermined time zone, a time zone in which the amount of activity of the human body 106 is stable is adopted. Can be adopted.

  In this case, the data classification unit 305 may classify only measurement data measured in a predetermined time zone. As a result, the reference value is calculated from the measurement data in a predetermined time zone, and changes in biological information can be detected with high accuracy.

  Alternatively, the data classification unit 305 may classify the measurement data in consideration of the time zone in addition to the posture, state, and distance. In this case, it is possible to accurately detect a change in biological information in consideration of the amount of activity of the human body 106 that varies depending on the time zone.

(Modification 2)
FIG. 13 is a diagram illustrating an outline of a biological information sensing device to which the information processing device according to the second modification of the present disclosure is applied. In Embodiment 1, as shown in FIG. 1, the sensing device 101 is configured separately from the air conditioner 105. However, in Modification 2, as shown in FIG. It is built in the air conditioner 105B. In the example of FIG. 13, the sensing device 101_B is disposed on the front surface of the air conditioner 105B so that the sensor surface faces the human body 106.

  In the second modification, the information processing apparatus 102 may be configured with, for example, a cloud server or a locally installed server.

(Modification 3)
FIG. 14 is a diagram illustrating a functional configuration of the biological information sensing device to which the information processing device according to the third modification of the present disclosure is applied. In the first embodiment, the information processing unit 102A is configured separately from the sensing unit 101A. However, in Modification 3, the sensing unit 101B includes the sensing unit 101A illustrated in FIG. 3 as illustrated in FIG. The information processing unit 102A is integrated. In Modification 3, an information processing apparatus is configured by the sensing unit 101B.

  In Modification 3, since the sensing unit 101A and the information processing unit 102A are integrated, the transmission unit 302 and the reception unit 303 existing in FIG. 3 are omitted. In Modification 3, the sensing unit 101B may be configured by a sensor unit built in the air conditioner 105, or may be configured by a dedicated sensor unit provided separately from the air conditioner 105. Also good.

  FIG. 15 is a diagram illustrating an example of a connection configuration of the biological information sensing device to which the information processing device according to the third modification of the present disclosure is applied. The sensing device 101_B corresponds to the sensing unit 101B illustrated in FIG. The sensing device 101_B is connected to the information display device 103 via the GW 104.

(Modification 4)
FIG. 16 is a diagram illustrating a functional configuration of the biological information sensing configuration to which the information processing apparatus according to the fourth modification of the present disclosure is applied. In the first embodiment, the time data providing unit 304 is provided in the information processing unit 102A. However, in the fourth modification, the time data providing unit 304 is provided in the sensing unit 101C. The time data adding unit 304 is provided between the sensor unit 301 and the transmission unit 302. On the other hand, the information processing unit 102C does not include the time data adding unit 304. In the modified example 4, since the process of giving the time data is performed by the sensing unit 101C, the processing load on the information processing unit 102C can be reduced. In the modified example 4, the sensing unit 101C is configured by a high function sensor built in the air conditioner 105 or a high function sensor configured separately from the air conditioner 105.

(Modification 5)
FIG. 17 is a diagram illustrating a functional configuration of a biological information sensing device to which an information processing device according to Modification 5 of the present disclosure is applied. In the example of FIG. 17, the information processing unit 102E is divided into an information processing unit 102D including a time data adding unit 304 and an information processing unit 102E.

  The information processing unit 102 </ b> D is configured by, for example, a high-functional sensor built in the air conditioner 105 or a home computer. The information processing unit 102E is configured by a cloud server, for example. The information processing unit 102D includes a receiving unit 303D for communicating with the sensing unit 101D and a transmitting unit 302D for communicating with the information processing unit 102E. The information processing unit 102E includes a receiving unit 303 for communicating with the information processing unit 102D. FIG. 18 is a diagram illustrating a first example of a connection configuration of a biological information sensing device to which an information processing device according to Modification 5 of the present disclosure is applied. The information processing device 102_D corresponding to the information processing unit 102D is connected to the information processing device 102_E and the information display device 103 corresponding to the information processing unit 102E via the GW 104. The information processing apparatus 102_D is connected to a sensing apparatus 101_D corresponding to the sensing unit 101D.

  FIG. 19 is a diagram illustrating a second example of the connection configuration of the biological information sensing device to which the information processing device according to the fifth modification of the present disclosure is applied. In the second example, the sensing device 101_D, the information processing device 102_D, the information processing device 102_E, and the information display device 103 are connected via the GW 104, respectively.

(Modification 6)
FIG. 20 is a diagram illustrating a connection configuration of a biological information sensing device to which an information processing device according to Modification 6 of the present disclosure is applied. In Modification 6, the GW 104 is connected to the information processing apparatus 102 and the information display apparatus 103 via the public communication network 801. Note that the sensing device 101 is connected to the GW 104.

(Modification 7)
FIG. 21 is a diagram illustrating a connection configuration of a biological information sensing device to which an information processing device according to Modification 7 of the present disclosure is applied. The modified example 7 is configured by using the public communication network 801 to configure the biological information sensing device according to the modified example 5. The GW 104 is connected to the information processing apparatus 102_E and the information display apparatus 103 via the public communication network 801. The sensing device 101 is connected to the GW 104 via the information processing device 102_D.

(Modification 8)
FIG. 22 is a diagram illustrating a connection configuration of the biological information sensing device to which the information processing device according to the modification 8 of the present disclosure is applied. In the eighth modification, the biological information sensing device according to the fifth modification is configured using the public communication network 801.

  The GW 104 is connected to the information processing apparatus 102_E and the information display apparatus 103 via the public communication network 801. The sensing device 101 and the information processing device 102_D are each connected to the GW 104 via a LAN.

(Modification 9)
FIG. 23 is a diagram illustrating an outline of a biological information sensing device to which the information processing device according to the ninth modification of the present disclosure is applied. Modification 9 is a living body information sensing device according to the second embodiment in which a sensing device 901 </ b> _A is disposed below the bedding 107. Even when the sensing device 901 </ b> _A is arranged below the bedding 107, the activity amount data can be measured as in the second embodiment.

(Modification 10)
In the second embodiment, the body motion value, the heart rate, and the pulse rate are employed as the activity amount, but at least one of these may be employed as the activity amount.

(Modification 11)
In Embodiment 1, the change detection unit 307 has been described as generating alerting information indicating an abnormality in the body temperature of the human body 106, but the present disclosure is not limited thereto. For example, the change detection unit 307 generates a difference between the body temperature of the human body 106 calculated from the thermal image data and the reference value as a comparison result, and transmits information indicating the difference to the information display unit 103A via the transmission unit 308. May be.

(Modification 12)
In the second embodiment, the change detection unit 907 has been described as generating alerting information indicating an activity amount abnormality of the human body 106, but the present disclosure is not limited thereto. The change detection unit 907 generates a difference between the activity amount of the human body 106 calculated from the activity amount data and the reference value as a comparison result, and transmits information indicating the difference to the information display unit 103A via the transmission unit 308. May be. Since the activity amount data includes the body motion value, the respiratory rate, and the heart rate, the change detection unit 907 includes the difference from the reference value for each of the body motion value, the respiratory rate, and the heart rate in the comparison result. You can do it.

  Modifications 1 to 8 are applicable to both Embodiments 1 and 2.

  The information processing apparatus and method according to each embodiment of the present disclosure can acquire biological information in a non-invasive manner with respect to the human body, and can appropriately grasp the biological state even when the posture of the measurement subject changes. By being able to do so, it is useful for daily biometric information recording and change detection in elderly care, and for controlling an air conditioner linked to a body movement state.

101A sensing unit 102A information processing unit 103A information display unit 106 human body 301 sensor unit 302 transmission unit 303 reception unit 304 time data addition unit 305 data classification unit 306 data management unit 307 change detection unit 308 transmission unit 309 reception unit 310 display unit 311 DB Unit 901A sensing unit 901 sensor unit 902A information processing unit 905 data classification unit 906 data management unit 907 change detection unit 911 DB unit

Claims (11)

Using a computer, obtain measurement data for calculating the biological information of the measurement subject from a measurement unit that performs measurement without contact,
Based on the content of the acquired measurement data, the measurement data is classified into a group corresponding to at least the posture of the measurement subject, the biological information is calculated from the classified measurement data, and the calculated biological information To the reference value corresponding to the group to which the measurement data belongs,
Causing the information presentation unit to make a notification based on the comparison;
Information processing method. In the classification of the measurement data, the measurement data is classified into the group using a machine learning model in which the measurement data is classified into the group based on the content of the measurement data.
The reference value is a representative value of the biological information of the group calculated from the measurement data classified into the group in the machine learning.
The information processing method according to claim 1. The information processing method further includes:
Classifying the measurement data into the group based on the content of the measurement data in the machine learning model,
In the machine learning, a representative value of the biological information of the group is calculated as the reference value of the group from the measurement data classified into the group.
The information processing method according to claim 2. The measurement unit is a thermal image sensor,
The biological information is body temperature,
The measurement data is thermal image data obtained by the thermal image sensor,
In the classification of the measurement data, the thermal image data is classified into the group based on the characteristics of the region representing the measurement subject included in the thermal image data obtained by the thermal image sensor,
In the comparison, the body temperature of the measurement subject is calculated from the classified thermal image data, and the calculated body temperature is compared with a reference value corresponding to a group to which the classified thermal image data belongs.
The information processing method according to claim 1. The measurement unit is a radio wave sensor,
The biological information is an activity amount including at least one of a body motion value, a respiratory rate, and a heart rate of the measurement subject,
The measurement data is activity amount data indicating the activity amount obtained by the radio wave sensor,
In the classification of the measurement data, the activity data is classified into the group based on waveform information of the activity data obtained by the radio wave sensor,
In the comparison, the activity amount of the measurement subject is calculated from the classified activity amount data, and the calculated activity amount is compared with a reference value corresponding to a group to which the classified activity amount data belongs. To
The information processing method according to claim 1. When the number of the measurement data that does not fall into any of the groups exceeds a reference number, information that prompts relearning of the machine learning model is generated.
The information processing method according to claim 3. The information processing method further acquires time data indicating a measurement time for the measurement data,
In the comparison, it is determined whether or not the measurement data is measured in a predetermined time zone based on the time data. When it is determined that the measurement data is measured in the predetermined time zone, the comparison is performed. Run,
The information processing method according to claim 1. The posture is the body direction of the measurement subject.
The information processing method according to claim 1. The notification based on the comparison is a comparison result between the biological information and the reference value, or an abnormality of the measurement subject.
The information processing method according to claim 1. An acquisition unit that acquires measurement data for calculating biological information of a measurement subject from a measurement unit that performs measurement without contact;
Based on the content of the received measurement data, a data classification unit that classifies the measurement data into a group according to at least the posture of the measurement target person;
A comparator that calculates the biological information from the classified measurement data, and compares the calculated biological information with a reference value corresponding to a group to which the measurement data belongs;
A transmitting unit that transmits information for causing the information presenting unit to perform notification based on the comparison to the information presenting unit;
An information processing apparatus comprising:   A program for causing a computer to execute the information processing method according to claim 1.

Images