JP2019145057A - Method for predicting health age - Google Patents

Abstract

To provide a numerical value prediction device related to a health age by utilizing artificial intelligence technology.SOLUTION: A numerical value prediction device related to a health age includes: a step for receiving expectation life expectancy learning data including learning physical condition information and learning expectation life expectancy information; a step for executing deep learning using the expectation life expectancy learning data to generate an expectation life expectancy model; and a step for receiving physical condition information, processing the received physical condition information with the expectation life expectancy prediction model and outputting a prediction value of expectation life expectancy.SELECTED DRAWING: Figure 3

Description

  The present disclosure relates to artificial intelligence technology, and more specifically, relates to prediction using artificial intelligence technology.

  Much research has been carried out so that the machine can change its operation according to the situation without human instruction, which led to the development of artificial intelligence. Among various artificial intelligence methods, artificial neural network technology that mimics a neural network requires a large amount of data learning and computing ability to achieve a satisfactory level of performance. For this reason, artificial neural network technology lacks industrial effectiveness because the cost required for learning is too high to use artificial neural network technology in the industry compared to the expected convenience. Although it has been recognized that the computer performance has dramatically improved and the learning method has been optimized, the cost of learning has been reduced, and artificial neural network technology has attracted attention again. . For these reasons, there are attempts to industrially use artificial neural network technology in various industrial fields, and there is also a need to use artificial neural network technology industrially in the healthcare field.

Korean Patent No. 10-1744775

  The present disclosure has been devised from the background art described above, and is intended to provide a prediction of numerical values related to health.

  A method for predicting a numerical value related to health is disclosed based on the first embodiment of the present disclosure for solving the above-described problems. The above-described method is a step of receiving expected life expectancy learning data including physical condition information for learning and expected life expectancy information for learning; performing deep learning using the above-mentioned expected life expectancy learning data, and an expected life expectancy prediction model Receiving physical condition information, processing the received physical condition information with the above-described expected life expectancy model, and outputting a predicted value of expected life expectancy.

  As an alternative, at least one of the predicted value of the age of aging and the predicted value of the health age is calculated by comparing the output predicted life expectancy value described above and the statistical expectation life information corresponding to the physical condition information described above. A step can further be included.

  As an alternative, the above-mentioned physical condition information and physical condition information for learning include name, personal ID, birth year, age, gender, personal information, health checkup date, height, weight, BMI, blood test values, ecological data, family calendar It may include at least one of alcohol consumption, exercise, smoking, medical date, medical institution information, medical breakdown, life / death, death date, and disease code.

  As an alternative, the method may further include the step of labeling the above-described expected life expectancy information for learning in each of the above-described physical condition information for learning of the above-mentioned expected life expectancy learning data.

  As an alternative, the step of labeling the above-mentioned expected life expectancy information for each of the above-mentioned learning physical condition information of the above-mentioned expected life expectancy learning data is performed when the subject included in the above-described learning physical condition information is a dead person, The difference between the information related to the time included in the physical condition information for learning and the information related to the death time of the target person is set as the expected life expectancy information, and is labeled in the physical condition information for learning; When the target person included in the physical condition information is not a dead person, the above-mentioned statistical life expectancy information based on the personal information of the target person is set as the expected life expectancy information, and the step of labeling the physical condition information for learning is performed. Can be included.

  As an alternative, the step of executing deep learning using the above-mentioned expected life expectancy learning data and generating the expected life expectancy model is the step of inputting the above-mentioned learning physical condition information into the artificial neural network; Deriving an error by comparing the output related to the physical condition information for learning output from the neural network with the expected life expectancy for learning; and the error derived as a result of the comparison to the artificial neural network Back propagation and updating the weight value of the artificial neural network may be included.

  As an alternative, the step of performing deep learning using the above-described learning data and generating the expected life expectancy model includes a learning epoch less than a predetermined epoch for learning the artificial neural network. The learning epoch for learning the artificial neural network is greater than or equal to the predetermined epoch by setting the learning rate of the artificial neural network to a predetermined value or higher. In this case, the learning rate of the aforementioned artificial neural network can be set to a predetermined value or less.

  As an alternative, receiving the aging learning data including the learning physical condition information, the expected life expectancy value and the learning aging information; deep learning using the aging learning data described above Executing the generation of an aging prediction model; and receiving the physical condition information, processing the received physical condition information with the aging prediction model and outputting a predicted value of aging. Can do.

  As an alternative, the step of generating deep learning using the above-mentioned aging years learning data to generate a numerical model for predicting the aging years related to health may include the above-described physical condition information for learning to the aging years prediction artificial neural network. A step of inputting; a step of deriving an error by comparing the output related to the physical condition information for learning output from the artificial neural network for predicting the age of aging and the information regarding the age of learning for aging described above; and as a result of the comparison The derived error may be back-propagated to the aging prediction artificial neural network to update the weight value of the artificial neural network.

  As an alternative, the method may further include the step of labeling the learning age information for each of the learning physical condition information of the learning age learning data.

  As an alternative, the step of labeling the learning age information for each of the learning physical condition information in the learning age learning data described above is performed when the subject included in the learning physical condition information is a dead person, The difference between the information about the time included in the physical condition information for learning and the information about the death time of the subject is used as age information, and is labeled in the physical condition information for learning; If the included subject is not a dead person, inputting the learning physical condition information into the artificial neural network before updating; and outputting the expected life expectancy prediction model related to the learning physical condition information Can be included in the above-mentioned aging years information and labeled with the above-described physical condition information for learning.

  As an alternative, the output of the expected life expectancy model related to the above-described learning physical condition information may be used as the above-mentioned aging year information and labeled to the above-described learning physical condition information.

  As an alternative, at least one of the above-mentioned physical condition information for learning is at least one of the birth year, age, sex, height, weight, BMI, blood test values, ecological data, family calendar, alcohol consumption, exercise, smoking, and disease code The method may further include grouping according to a predetermined criterion including:

  As an alternative, after performing deep learning using the expected life expectancy learning data grouped according to the above-mentioned predetermined criteria, the step of generating the grouped expected life expectancy prediction model by group is performed. Further can be included.

  As an alternative, the physical condition information for learning described above is a re-learning life expectancy information obtained by processing the physical condition information for re-learning that the target person is a dead person and the target person is a non-dead person with the generated expected life expectancy prediction model. Generating the re-learning data by labeling the re-learning expected life expectancy information for each of the re-learning physical condition information; using the expected remaining life learning data and the re-learning data described above And performing the deep learning to update the generated expected life expectancy prediction model.

  In a second embodiment of the present disclosure, an apparatus for predicting numerical values related to health is disclosed. The above-mentioned apparatus receives a life expectancy learning data including physical condition information for learning and life expectancy information for learning; a learning portion that generates a life expectancy prediction model by executing deep learning using the life expectancy learning data described above And a health-related prediction model that receives physical condition information, processes the received physical condition information with the above-described expected life expectancy model, and outputs a predicted value of expected life expectancy.

  In a third embodiment of the present disclosure, a computer program stored on a computer readable storage medium containing encoded instructions is disclosed. When the computer program is executed by one or more processors of a computer system, the computer program executes the following steps for causing the one or more processors to predict health figures, The steps are: receiving expected life expectancy learning data including physical condition information for learning and expected life expectancy information for learning; performing deep learning using the aforementioned expected life expectancy learning data to generate an expected life expectancy prediction model And receiving physical condition information, processing the received physical condition information with the above-described expected life expectancy model, and outputting a predicted value of expected life expectancy.

  The present disclosure can provide a numerical prediction of health.

It is a block diagram of the numerical prediction apparatus (100) which concerns on the health in 1st Example of this indication. It is a lineblock diagram of a neural network in the 1st example of this indication. It is a flowchart of the numerical prediction method which concerns on the health in 1st Example of this indication. It is a flowchart of the method of labeling learning life expectancy information in each physical condition information for learning in the 1st example of this indication. It is a flowchart of the method of calculating | requiring the predicted value of aging years using the aging years learning data in 1st Example of this indication. It is a flowchart of the method of producing | generating the expected life expectancy model by a reference | standard by grouping the physical condition information for learning by a predetermined reference | standard. It is a flowchart of the method of receiving the data for re-learning in the model of expected life expectancy, and performing deep learning again and updating an expected life expectancy model. 2 is a block diagram illustrating means for embodying a method for predicting a numerical value related to health according to a first embodiment of the present disclosure; 2 is a block diagram illustrating a module for implementing a method for predicting a health value according to a first embodiment of the present disclosure; 2 is a block diagram illustrating logic for embodying a method for predicting a numerical value related to health according to a first embodiment of the present disclosure; 2 is a block diagram illustrating a circuit for embodying a method for predicting a numerical value related to health according to a first embodiment of the present disclosure; FIG. FIG. 2 is a simplified, general schematic diagram of an exemplary computing environment in which embodiments of the present disclosure can be implemented.

  Various embodiments are described below with reference to the drawings. In this specification, various descriptions are given to enhance the understanding of the present disclosure. However, it will be apparent that such embodiments can be practiced without such specific explanation. In other instances, well-known structures and devices are provided in block diagrams to facilitate description of the embodiments.

  As used herein, the terms “component”, “module”, “system”, etc. refer to computer-related entities, hardware, firmware, software, software and hardware combinations, or execution of software. For example, a component can be, but is not limited to being, a process of processing to be executed on a processor, a processor, an object, an execution thread, a program, and / or a computer. For example, an application running on a computing device and the computing device can all be components. One or more components can reside in a processor and / or thread of execution, and one component can be localized within a single computer or distributed to two or more multiple computers It is also possible. Such components can also be executed from a variety of computer readable media having a variety of data structures stored therein. A component may be, for example, a signal having one or more data packets (eg, data and / or signals from one component interacting with another component in a local system, distributed system, and other systems and networks such as the Internet. Can be communicated through local and / or remote processing.

  In addition, the terms “include” and / or “include” include the characteristics and / or configuration of what is included or included, both when used as a predicate and when modifying a noun or the like from the back. It should be understood that an element is present but does not exclude the presence or addition of one or more other features, components and / or groups thereof. In addition, in the present specification and claims, the singular generally means “one or more” when the number is not specified or cannot be clearly read in the context of the singular form. Should be construed to do.

  The description of the illustrated embodiments is provided to enable any person skilled in the art to use or practice the present disclosure. Various modifications to such embodiments will be apparent to those of ordinary skill in the art of the present disclosure, and the general principles defined herein may be practiced without departing from the scope of the present disclosure. This embodiment can be applied. Accordingly, the present disclosure is not limited to the embodiments shown herein but should be analyzed in the broadest sense consistent with the principles and novel features shown herein.

  FIG. 1 is a block diagram of a numerical prediction device (100) related to health in the first embodiment of the present disclosure. The numerical value prediction apparatus (100) related to health in the first embodiment of the present disclosure includes a receiving unit (110), a health-related prediction model (120), a labeling unit (130), a learning unit (140), and a calculation unit (150). Can be included. The block diagram shown in FIG. 1 shows the numerical prediction apparatus (100) related to health in a simplified configuration, and the present disclosure is not limited to this, and further includes other components necessary for driving. Can be included.

  The receiving unit (110) can receive data and the like for executing the numerical prediction method relating to health in the first embodiment of the present disclosure from other computing devices, servers, and the like. The receiving unit (110) can receive life expectancy learning data including physical condition information for learning and life expectancy information for learning. Physical condition information for learning includes name, personal ID, year of birth, age, gender, personal information, health checkup date, height, weight, BMI, blood test values, ecological data, family calendar, drinking, exercise, smoking , Medical date, medical institution information, medical breakdown, life / death, death date, and disease code. The physical condition information for learning can include any information related to the current state of the subject and the current health state for learning the numerical prediction model related to health in the first embodiment of the present disclosure. The expected life expectancy information for learning can include arbitrary information related to the expected life expectancy of the subject. The expected life expectancy information for learning is the time interval between the subject's physical condition information (that is, the subject's health check information) and the subject's death time (that is, When the subject is a dead person, a time interval between when the physical condition information for learning is generated and when the subject died can be included. Expected life expectancy information for learning is information on the expected life expectancy of the subject determined by statistics when the subject is not a dead person (that is, if the subject of the physical condition information for learning is not a dead but a man in their 30s According to the above, the expected life expectancy information can be 50 years if survival is predicted until age 80).

  The receiving unit (110) can receive physical condition information. Physical condition information includes name, personal ID, year of birth, age, gender, personal information, health checkup date, height, weight, BMI, blood test values, ecological data, family calendar, drinking, exercise, smoking, medical treatment It includes at least one of date, medical institution information, medical breakdown, life / death, death date, and disease code. The physical condition information includes an arbitrary information related to the current state of the subject and the current health state to be transmitted to the numerical prediction model related to health in the first embodiment of the present disclosure and to output the numerical predicted value related to health. Can do. Numerical predictive values related to health include predicted values of expected life expectancy, predicted values of age of aging, and the like. Expected life expectancy indicates how many years the subject is expected to live. For example, if the expected life expectancy is 30, it is expected to be alive for the next 30 years. The age of aging represents how much physical aging has progressed compared to the actual age of the subject. For example, when the predicted value of age is -1, physical aging has not progressed by one year compared to the actual age. When the predicted value of age is 1, physical aging progresses one year earlier than the actual age. Can be out.

  The receiving unit (110) can receive aging years learning data including predicted life expectancy and learning aging information. The predicted value of the expected life expectancy may be a value related to the expected life expectancy predicted by the expected life expectancy prediction model generated in the first embodiment of the present disclosure. Aging age information for learning can be a difference between a predicted value of expected life expectancy for a subject and statistical expected life expectancy information. For example, if the predicted life expectancy of a subject who is a 30 year old male is 40 years and the statistical life expectancy is 50 years, the subject has a life expectancy shorter than the level generally expected, It can be judged that aging is progressing more than my real age. In this case, the age of the subject who is a man in his 30s can be 10 years. Conversely, if the expected life expectancy is 60 years and the statistical life expectancy is 50 years, the subject has a longer expected life expectancy than the expected level, so the aging has not progressed more than his actual age. It can be judged. In this case, the age of the subject who is a man in his 30s can be -10 years. For example, the receiving unit 110 can receive an electronic health record (EHR) from a hospital server or government server, and can receive an electronic medical record (EMR) from a hospital server or the like. Can be received. The receiving unit (110) can receive a user's health check record and the like from a user terminal (not shown). In addition, the receiving unit (110) can receive a numerical prediction value related to health from an existing model as new learning data. The above description is merely illustrative and the present disclosure is not limited thereto.

  The health-related prediction model (120) may include an artificial neural network that outputs numerical prediction values related to health. Numerical predictive values related to health include predicted values of expected life expectancy, predicted values of age of aging, and the like. The health-related prediction model (120) inputs the physical condition information received from the receiving unit (110) to the artificial neural network, and outputs the numerical prediction value related to health output to the user, thereby outputting the first of the present disclosure. Numerical predictions related to health in the embodiment can be executed. The above description is merely an example, and the present disclosure is not limited thereto. The health-related prediction model (120) may include an expected life expectancy model and an aging age prediction model by artificial neural network learning of the learning unit (140). The health-related prediction model (120) includes an artificial neural network, and can output a predicted value of expected life expectancy and a predicted value of age of aging.

  The health-related prediction model (120) may include a life expectancy prediction model and an aging prediction model.

  The labeling unit (130) can associate learning data with a label for supervisory learning. The labeling unit (130) can label the expected life expectancy information for learning to each of the physical condition information for learning of the expected life expectancy learning data. For example, the labeling unit (130) can label the learning physical condition information related to the subject named A with the expected life expectancy information for the subject named A.

  When the subject included in the physical condition information for learning is a dead person, the labeling unit (130) determines the difference between the information related to the time included in the physical condition information for learning and the information related to the death time of the subject. The expected life expectancy information can be labeled on the above-described physical condition information for learning. When the subject included in the physical condition information for learning is not a dead person, the labeling unit (130) sets the expected life expectancy information based on the personal information of the subject as the above expected life expectancy information, and performs the above learning You can label your physical condition information. For example, the labeling unit (130) determines the difference between the time information included in the physical condition information for learning of the subject A who is the subject A (ie, the health check date) and the time when the subject A actually died. (That is, the time from the health check date to the death date) can be used as life expectancy information, and can be labeled on the physical condition information for learning of the subject A. If the target person B is not a dead person, the labeling unit (130) uses the expected life expectancy information based on the personal information of the target person B (for example, the expected life expectancy based on the expected life expectancy information by age of the Statistics Agency) It is possible to label B's physical condition information for learning. For example, if the subject B is a man in his 30s and has a statistical life expectancy of 50 years, the expected life expectancy of 50 years can be labeled on the physical condition information for learning of the subject B. Further, the labeling unit (130) can label the expected life expectancy information due to the disease into the physical condition information for learning of the subject C when the subject C is not a dead person but a patient with a very high probability of dying in the future. For example, when the target person C is a patient in the 50s who is in the 4th stage of lung cancer and the average life of the 4th stage of lung cancer is 5 years, the expected life expectancy of 5 years can be labeled in the physical condition information for learning C.

  The labeling unit (130) can label the learning age information for each of the learning physical condition information of the age learning data described above. When the subject included in the physical condition information for learning is a dead person, the labeling unit (130) determines the difference between the information related to the time included in the physical condition information for learning described above and the information related to the death time of the target person described above. It can be labeled with the above-mentioned physical condition information for learning as age information. If the subject included in the learning physical condition information is not a dead person, the labeling unit (130) inputs the learning physical condition information into the generated expected life expectancy model, and the learning physical condition information described above. The output of the above life expectancy prediction model according to the above can be used as the above-mentioned age information, and can be labeled with the above-described physical condition information for learning.

  The labeling unit (130) uses the above-mentioned physical condition information for learning among the birth year, age, sex, height, weight, BMI, blood test values, ecological data, family calendar, alcohol consumption, exercise activity, smoking activity, and disease code. They can be grouped according to predetermined criteria, including at least one. For example, even in the case of physical condition information for learning with the same 30-year-old male as a subject, the statistical life expectancy may differ depending on the subject's birth year, income level, and the like. Accordingly, the labeling unit 130 can group the learning data according to a predetermined criterion so that different health-related prediction models are generated for each group.

  When the subject of the physical condition information for learning is a non-dead person, the labeling unit (130) uses the predicted life expectancy predicted by the generated expected life expectancy model (that is, expected life expectancy information for re-learning) Can be labeled with physical condition information for learning that is non-dead. That is, in the case of physical condition information for learning where the target person is a non-dead, the labeling unit (130) can also label the corresponding physical condition information for learning using statistical expected life expectancy information, It is also possible to label predicted values of expected life expectancy using an expected life expectancy model. The labeling unit (130) can generate re-learning data by labeling the expected life expectancy information for re-learning for each physical condition information for re-learning.

  The learning unit (140) may perform deep learning on the artificial neural network of the health-related prediction model (120) using expected life expectancy learning data to generate an expectation life expectancy prediction model. The learning unit (140) inputs the learning physical condition information to the artificial neural network of the health-related prediction model (120) and outputs the learning physical condition information output from the artificial neural network and the above-described learning physical condition information. Health-related prediction by comparing the expected life expectancy for learning and deriving the error and back-propagating the error derived as a result of the comparison to the artificial neural network and updating the weight value of the artificial neural network The model (120) can be trained to output a predicted value of expected life expectancy. Thereby, the learning unit (140) can generate an expected life expectancy model. The expected life expectancy prediction model of the present disclosure can output the expected life expectancy of the subject when the physical condition information of the subject is input. In the present disclosure, the artificial neural network can be configured by an RNN (recurrent neural network), and can also be configured by an LSTM (Long Short Term Memory) in which the problem of long-term memory is solved in the RNN.

  When the learning epoch for learning the artificial neural network is equal to or less than a predetermined epoch, the learning unit (140) sets the learning rate of the artificial neural network to a predetermined numerical value. If the learning epoch for learning the above-described artificial neural network is greater than or equal to a predetermined epoch, the learning rate of the artificial neural network may be set to a predetermined value or less. it can. For example, in order to learn the artificial neural network, the learning unit (140) sets the learning rate of 0.8 in the initial 100,000 epochs and increases the initial learning speed to verify the effectiveness of the artificial neural network model. In addition, the learning result can be finely adjusted by setting a learning rate of 0.2 for 100,000 or more epochs.

  The learning unit 140 may perform deep learning on the artificial neural network of the health-related prediction model 120 using the aging years learning data to generate an aging year prediction model. The learning unit (140) can use the artificial neural network of the health-related prediction model (120) to learn so that the health-related prediction model (120) outputs a predicted value of age. The learning unit (140) may perform deep learning using the aging years learning data to train the health-related prediction model (120) to output the aging years prediction data. The learning unit (140) can input the physical condition information for learning into the aging age prediction artificial neural network of the health related prediction model (120). The learning unit (140) derives an error by comparing the learning physical condition information output from the learning physical condition information output from the artificial aging prediction artificial neural network with the learning age information, and predicts the derived error based on the aging time. The health-related prediction model 120 can learn to output the predicted value of the aging time by backpropagating to the artificial neural network and updating the weight value of the aging age prediction artificial neural network. Thereby, the learning unit (140) can generate an aging years prediction model.

  The learning unit (140) performs the deep learning using the expected life expectancy learning data grouped according to the criteria determined in advance by the labeling unit (130), and then performs the grouped expected life expectancy by group. A predictive model can be generated. In other words, the learning part (140) is grouped by social factors because the expected life expectancy may be different due to social influences such as income level, etc. even if it is a subject in the same current state. By making use of the expected life expectancy learning data to generate a life expectancy prediction model, social factors can be reflected.

  The learning unit (140) can update the expected life expectancy prediction model generated by performing deep learning using the expected life expectancy learning data and the relearning data. The updated life expectancy prediction model can reduce the impact of statistical life expectancy information by using re-learning data that is labeled using the life expectancy prediction model for non-dead physical condition information . Accordingly, it is possible to prevent errors that may occur due to statistical expected life expectancy information that may change due to social factors such as the times by updating the expected life expectancy prediction model.

  The calculation unit (150) can receive a predicted value of expected life expectancy, a predicted value of age of aging, and physical condition information in the receiving unit (110) and the expected life expectancy-related prediction model (120).

  The calculation unit (150) can generate a numerical predicted value related to new health from statistical expected life expectancy information of physical condition information, a predicted value of expected life expectancy, and physical condition information. The numerical prediction value relating to the new health can be, for example, a prediction value of the health age. The health age includes the meaning of the subject's physical health age. A numerical prediction value relating to a new health can be obtained through a mathematical operation based on a mathematical relationship between the existing prediction value and the numerical prediction value relating to a new health. For example, the predicted value of healthy age can be obtained by adding the statistical expected life expectancy information to the age of the subject included in the physical condition information and subtracting the predicted value of expected life expectancy. Moreover, the predicted value of healthy age can be obtained by adding the age of the subject to the age of the subject included in the physical condition information.

  The calculation unit (150) compares the output expected life expectancy value described above with the statistical expectation life information corresponding to the physical condition information, and at least one of the predicted age value and the predicted health age value. Can be calculated. The predicted value of health age is the difference between the age of the subject included in the physical condition information and the predicted value of age. For example, when the age of the subject is 30 years old and the predicted value of the age of aging is +10, the health age can be 40 years old because the subject is aging more than their own age. The predicted value of the age of aging can be obtained by subtracting the predicted value of expected life expectancy from statistical expected life expectancy information.

  FIG. 2 is a configuration diagram of the neural network in the first embodiment of the present disclosure.

  Throughout this specification, neural networks, network functions, and neural networks can be used interchangeably. A neural network can be composed of a collection of a plurality of computation units connected to each other, which can be generally referred to as “nodes”. Such a “node” may also be referred to as a “neuron”. The neural network includes at least one node. Nodes (or neurons) constituting a neural network can be connected to each other by one or more “links”.

  In a neural network, one or more nodes connected through a link can form an input node and an output node relationship with each other. The concept of input node and output node is relative, and any node that is an output node for one node can be an input node in relation to other nodes, and vice versa. It is possible. As described above, the input node-to-output node relationship can be created around a link. One or more output nodes can be connected to one input node through a link, and vice versa.

  In the relationship between the input node and the output node connected through one link, the value of the output node may be determined based on data input to the input node. Here, the node that interconnects the input node and the output node can have a weight. The weights can be variable and can be changed by the user or algorithm to perform the necessary functions for the neural network. For example, when one or more input nodes are connected to one output node by each link, the output node is a value input to the input node connected to the output node and a link corresponding to each input node. The output node value can be determined based on the weighting value set in (1).

  As described above, in the neural network, one or more nodes are interconnected through one or more links to form an input node and an output node relationship in the neural network. Depending on the number of nodes and links in the neural network, the relationship between the nodes and links, and the weight value assigned to each of the plurality of links, the characteristics of the neural network may be determined. For example, if there are two neural networks having the same number of nodes and links and different weight values between the links, the two neural networks can be recognized as different from each other.

  As shown in FIG. 2, a neural network can include one or more nodes. Some of the nodes constituting the neural network can form a layer based on the distance from the first input node. For example, a node whose distance is n from the first input node Can be made up of n layers. The distance from the first input node can be defined by the minimum number of links that must be traversed to reach the appropriate node from the first input node. However, the definition of such a layer is arbitrarily taken up for explanation, and the number of layers in the neural network can be defined in a manner different from that described above. For example, the layer of nodes can be defined by the distance from the final output node.

  The first input node may mean one or more nodes in which data is directly input without passing through links in relation to other nodes among nodes in the neural network. Or, in a neural network network, it can mean a node that does not have other input nodes connected by a link in the relationship between nodes based on the link. Similarly, the final output node may mean one or more nodes having no output node in relation to other nodes among the nodes in the neural network. The hidden node may mean a node constituting the neural network, not the first input node and the last output node. In FIG. 2, the output node is not shown. In the first embodiment of the present disclosure, the neural network can have more nodes in the input layer than nodes in the hidden layer close to the output layer, and the number of nodes decreases as the process proceeds from the input layer to the hidden layer. Can be a neural network.

  In the first embodiment of the present disclosure, the neural network may include a multi layer perceptron (MLP), a recurrent neural network (RNN), a regenerative neural network (CNN), and the like so as to be able to process data. In addition, in the first embodiment of the present disclosure, the neural network of the present disclosure is provided with an LSTM (long short) so as to prevent a decrease in performance due to a long range dependency varying gradient that may occur due to a long event. -Term memory). In the first embodiment of the present disclosure, stochastic gradient descent (SGD), Momentum, Adam, Adagrad, RMSprop, etc. can be used as an optimization technique for the artificial neural network method. In addition, if the learning data D can be learned only once, a parameter that minimizes the error function can be obtained through an epoch that is repeated several times. You can decide whether you can finish.

  Unlike a general feedforward neural network, the output of the hidden layer can be the input of the hidden layer. The RNN is a neural network having a memory capability by performing calculation while simultaneously considering current input data and data received in the past and having a feedback structure. Thus, the RNN can be trained in the data to analyze the current data according to the meaning of the previous data. LSTM, which is one of RNNs, is also called a long short term memory network, and can learn long-term dependence. In the first embodiment of the present disclosure, the neural network is not limited to LSTM, which is one of the RNNs, but may be a depth gated neural network (depth gated RNN), a clock device circulating neural network (clock work RNN), or the like. Any neural network capable of processing data can be included.

  The neural network (200) of FIG. 2 can include one or more hidden layers. The hidden node of the hidden layer can use the output of the previous layer and the output of the surrounding hidden nodes as inputs. The number of hidden nodes for each hidden layer may be the same or different. The number of nodes in the input layer can be determined based on the number of data fields in the input data, and may be the same as or different from the number of hidden nodes. Input data input to the input layer can be calculated by a hidden node of the hidden layer and can be output by a fully connected layer (FCL) which is an output layer.

  FIG. 3 is a flowchart of a numerical prediction method related to health in the first embodiment of the present disclosure.

  The health-related numerical prediction device (100) can receive expected life expectancy learning data (310). The expected life expectancy learning data can include physical condition information for learning and expected life expectancy information for learning. Physical condition information for learning includes name, personal ID, year of birth, age, gender, personal information, health checkup date, height, weight, BMI, blood test values, ecological data, family calendar, drinking, exercise, smoking , Medical date, medical institution information, medical breakdown, life / death, death date, and disease code. The physical condition information for learning can include any information related to the current state and the current health state of the subject for learning the numerical prediction model related to health in the first embodiment of the present disclosure. The expected life expectancy information for learning can include arbitrary information related to the expected life expectancy of the subject. The expected life expectancy information for learning is the time interval between the subject's physical condition information (that is, the subject's health check information) and the subject's death time (that is, When the subject is a dead person, a time interval between when the physical condition information for learning is generated and when the subject died can be included. Expected life expectancy information for learning is information on the expected life expectancy of the subject determined by statistics when the subject is not a dead person (that is, if the subject of the physical condition information for learning is not a dead but a man in their 30s According to the above, the expected life expectancy information can be 50 years if survival is predicted until age 80).

  The health-related numerical prediction apparatus (100) can generate a life expectancy prediction model by executing deep learning using the life expectancy learning data (320). The health-related numerical prediction device (100) can include an artificial neural network that outputs a predicted value of expected life expectancy. The health-related numerical prediction device (100) inputs the physical condition information for learning contained in the above-mentioned expected life expectancy learning data to the artificial neural network of the health-related numerical prediction device (100), and from the aforementioned artificial neural network. An error is derived by comparing the output related to the physical condition information for learning described above with the expected life expectancy for learning, and the error derived as a result of the comparison is propagated back to the artificial neural network to The weight value of the artificial neural network can be updated and the expected life expectancy model can be trained to output the expected life expectancy. Thereby, the numerical prediction apparatus (100) concerning health can generate an expected life expectancy model. In the present disclosure, the artificial neural network can be configured by an RNN (recurrent neural network), and can also be configured by an LSTM (Long Short Term Memory) in which the problem of long-term memory is solved in the RNN.

  After receiving physical condition information in the expected life expectancy model, the numerical prediction device (100) related to health can output a predicted value of expected life expectancy (330). The health-related numerical prediction device (100) can receive physical condition information. Physical condition information includes name, personal ID, year of birth, age, gender, personal information, health checkup date, height, weight, BMI, blood test values, ecological data, family calendar, drinking, exercise, smoking, medical treatment It includes at least one of date, medical institution information, medical breakdown, life / death, death date, and disease code. The physical condition information can include any information related to the current state of the subject and the current health state for outputting the predicted value of expected life expectancy to the numerical prediction model related to health in the first embodiment of the present disclosure. . The health-related numerical prediction device (100) inputs the received physical condition information to the expected life expectancy prediction model, and outputs the expected life expectancy value output to the user, thereby predicting the expected life expectancy in the first embodiment of the present disclosure. Can be executed.

  FIG. 4 is a flowchart of a method for labeling learning expected life expectancy information for each learning physical condition information in the first embodiment of the present disclosure.

  The health-related numerical prediction device (100) can perform labeling based on whether or not the physical condition information for learning is a dead person (410).

  If the physical condition information for learning is a dead person, the numerical predicting apparatus (100) relating to health indicates the difference between the information related to the time included in the physical condition information for learning and the information related to the death time of the target person as described above. And labeling to the above-described physical condition information for learning (421). The numerical prediction device (100) relating to health includes information on the time included in the physical condition information for learning of the subject A who is the subject A (ie, the health check date) and the time when the subject A actually died. Difference (that is, the time from the medical checkup date to the death date) can be used as life expectancy information and can be labeled on the physical condition information for learning of the subject A. For example, when the subject A is a 50-year-old male and died at age 54, the expected life expectancy of 4 years can be labeled on the physical condition information for learning of the subject A.

  The health-related numerical prediction device (100) uses the expected life expectancy information based on the personal information of the subject as the above-mentioned life expectancy information, if the physical condition information for learning is a non-dead person. It can be labeled (422). The statistical life expectancy information can be the above-mentioned year, gender, age-specific life table, life table from which a specific cause of death has been excluded, and the average number of years of life due to a specific disease. When the target person B is not a dead person, the numerical prediction device (100) related to health is expected life expectancy information based on personal information of the subject person B (for example, expected life expectancy based on expected life expectancy information by age of the Statistics Agency) ) To the physical condition information for learning of the subject B. For example, if the subject B is a man in his 30s and has a statistical life expectancy of 50 years, the expected life expectancy of 50 years can be labeled on the physical condition information for learning of the subject B.

  FIG. 5 is a flowchart of a method for obtaining a predicted value of aging years using aging years learning data in the first embodiment of the present disclosure.

  The health-related numerical prediction device (100) can receive aging age learning data including predicted life expectancy and physical condition information (510). The predicted value of the expected life expectancy may be a value related to the expected life expectancy predicted by the expected life expectancy prediction model generated in the first embodiment of the present disclosure. Aging age information for learning can be a difference between a predicted value of expected life expectancy for a subject and statistical expected life expectancy information. For example, if the predicted life expectancy of a subject who is a 30 year old male is 40 years and the statistical life expectancy is 50 years, the subject has a life expectancy shorter than the level generally expected, It can be judged that aging is progressing more than my real age. In this case, the age of the subject who is a man in his 30s can be 10 years. Conversely, if the expected life expectancy is 60 years and the statistical life expectancy is 50 years, the expected life expectancy will be longer than the expected level, so the aging will progress faster than his actual age. It can be judged that it is not. In this case, the age of the subject who is a man in his 30s can be -10 years.

  The numerical prediction device (100) related to health can generate an aging prediction model after deep learning using aging learning data (520). The health-related numerical prediction device (100) can input the physical condition information for learning to the aging age prediction artificial neural network of the health-related numerical prediction device (100). The health-related numerical prediction apparatus (100) derives an error by comparing the output related to the physical condition information for learning output from the artificial neural network predicting the aging time with the information for the aging age for learning described above. Is propagated back to the aging prediction artificial neural network to update the weight of the aging prediction artificial neural network so that the health-related prediction model 120 outputs the aging years. As a result, the health-related numerical prediction device (100) can generate an aging years prediction model.

  After receiving the physical condition information in the aging prediction model, the numerical prediction device (100) related to health can output a prediction value of aging years (530). The health-related numerical prediction device (100) inputs the received physical condition information to the aging years prediction model, and outputs the predicted values of the aging years output to the user, thereby relating to the health according to the first embodiment of the present disclosure. Numerical predictions can be performed.

  FIG. 6 is a flowchart of a method for generating the expected life expectancy model based on the criteria by grouping the physical condition information for learning according to a predetermined criterion.

  The health prediction apparatus (100) can group physical condition information for learning according to a predetermined criterion (610). The predetermined criteria may include at least one of the year of birth, age, sex, height, weight, BMI, blood test values, ecological data, family calendar, alcohol consumption, exercise status, smoking status, and disease code. it can.

  The numerical prediction apparatus (100) related to health executes the deep learning using the expected life expectancy learning data grouped according to a predetermined criterion, and then performs the grouped expected life expectancy prediction model for each group. Can be generated (620). For example, the statistical life expectancy when a 1990-year-old is 30 years old and the statistical life expectancy when a 1970-year-old is 30 are different depending on social factors such as changes in the times. There can be. Therefore, the health-related numerical prediction device (100) groups the expected life expectancy learning data based on a predetermined social factor, and then performs learning using the grouped expected life expectancy learning data. A life expectancy prediction model for each group can be generated. By generating a prediction model using the grouped life expectancy learning data, the health-related numerical prediction apparatus (100) of the present disclosure can provide a prediction value of life expectancy reflecting social factors.

  FIG. 7 is a flowchart of a method for updating the expected life expectancy prediction model by receiving re-learning data in the expected life expectancy model and executing deep learning again.

  The health-related numerical prediction device (100) can receive expected life expectancy learning data in which the subject is a dead person (710).

  The numerical prediction apparatus (100) related to health can generate an expected life expectancy model after performing deep learning using expected life expectancy learning data whose subject is a dead person (720). The health-related numerical prediction device (100) inputs the physical condition information for learning that the above-mentioned subject is a dead person into the artificial neural network of the health-related prediction model (120) and outputs the information from the artificial neural network. The error is derived by comparing the output related to the physical condition information for learning that the subject of the death is the death with the expected life expectancy for learning, and the error derived as a result of the comparison is propagated back to the artificial neural network. It is possible to learn to output the expected life expectancy by updating the weight value of the aforementioned artificial neural network. Thereby, the numerical prediction apparatus (100) concerning health can generate an expected life expectancy model.

  The health-related numerical prediction device (100) can generate physical life expectancy information for re-learning by receiving physical condition information for re-learning whose target is a non-dead person in the life expectancy prediction model (730). Physical condition information for relearning includes name, personal ID, year of birth, age, gender, personal information, health checkup date, height, weight, BMI, blood test values, ecological data, family calendar, alcohol consumption, exercise, smoking It includes at least one of presence / absence, date of medical treatment, information on medical institution, medical breakdown, life / death, date of death, and disease code. The physical condition information for relearning can include any information related to the current state and current health state of the subject for learning the numerical prediction model related to health in the first embodiment of the present disclosure. The expected life expectancy information for relearning can be a value output from the numerical prediction device (100) related to health by inputting physical condition information for relearning to the numerical prediction device (100) related to health. That is, the numerical predicting apparatus (100) related to health according to the first embodiment of the present disclosure receives physical condition information of non-dead persons, and uses the learned health-related prediction model (120) to determine non-dead persons. Expected life expectancy information can be generated.

  The health-related numerical prediction device (100) can generate re-learning data by labeling the above-described expected life expectancy information for re-learning with each physical condition information for re-learning (740). The health-related numerical predicting apparatus (100) labels predicted values of expected life expectancy (that is, expected life expectancy information for re-learning) generated by receiving physical condition information for re-learning on each physical condition information for learning. Can do. The health-related numerical prediction device (100) can label predicted values of expected life expectancy using the generated expectation expectancy model on each physical condition information for re-learning.

  The health-related numerical prediction device (100) can update the expected life expectancy model by executing deep learning using the expected life expectancy learning data and the aforementioned relearning data (750). The numerical predicting apparatus (100) relating to health causes the above-mentioned object to input the physical condition information for learning and the physical condition information for relearning to the artificial neural network of the health-related prediction model (120) and output from the artificial neural network. An error is derived by comparing the output related to the physical condition information for learning and the physical condition information for re-learning with the expected life expectancy for learning and the expected life expectancy for re-learning, and the error derived as a result of the comparison described above It is possible to learn to output the expected life expectancy by backpropagating to the neural network and updating the weight value of the artificial neural network. As a result, the health-related numerical prediction device (100) can update the expected life expectancy model. The health prediction apparatus (100) according to the first embodiment of the present disclosure generates a life expectancy prediction model using data of a dead person, and extracts a predicted value of expected life expectancy from the data of a non-dead person through this. By executing the relearning, it is possible to reduce the dependence on the expected life expectancy information that may change with time, and provide a consistent result.

  FIG. 8 is a block diagram illustrating means for embodying a method for predicting a numerical value related to health according to the first embodiment of the present disclosure.

  Based on the first embodiment of the present disclosure, a method for predicting a numerical value related to health is a means for receiving expected life expectancy learning data including physical condition information for learning and expected life expectancy information for learning (1210); Means (1220) for generating a life expectancy prediction model by performing deep learning using life expectancy learning data; and receiving physical condition information, and processing the received physical condition information with the expected life expectancy prediction model This can be realized by means (1230) for outputting a predicted value of expected life expectancy.

  In the alternative embodiment, at least one of the predicted value of age of aging and the predicted value of healthy age is obtained by comparing the predicted value of expected life expectancy output and the statistical expectation information corresponding to the physical condition information described above. A means for calculating can be further included.

  In the embodiment as an alternative, the above-mentioned physical condition information and physical condition information for learning include name, personal ID, birth year, age, gender, personal information, health checkup date, height, weight, BMI, blood test value, ecological measurement data , Family calendar, presence / absence of drinking, presence / absence of exercise, presence / absence of smoking, medical date, medical institution information, medical breakdown, life / death, death date, and disease code.

  The alternative embodiment may further include means for labeling the learning expected life expectancy information in each of the learning physical condition information of the expected life expectancy learning data.

  In an alternative embodiment, the means for labeling the above-mentioned learning expected life expectancy information in each of the above-described learning life expectancy information of the above-mentioned expected life expectancy learning data is that the subject included in the above-described learning physical condition information is a dead person. In some cases, the difference between the information related to the time included in the physical condition information for learning and the information related to the death time of the subject is set as the expected life expectancy information, and is labeled with the physical condition information for learning; and If the subject included in the physical condition information for learning is not a dead person, the expected life expectancy information based on the personal information of the subject described above is used as the expected life expectancy information, and is labeled on the physical condition information for learning described above. Means may be included.

  In the alternative embodiment, the means for generating the expected life expectancy model by executing the deep learning using the expected life expectancy learning data is the means for inputting the learning physical condition information to the artificial neural network; Means for deriving an error by comparing the output related to the physical condition information for learning output from the neural network with the expected life expectancy for learning; and the error derived as a result of the comparison to the artificial neural network Means for backpropagating and updating the weight values of the artificial neural network described above may be included.

  In the alternative embodiment, the means for generating the expected life expectancy model by executing the deep learning using the learning data described above is a learning epoch for learning the artificial neural network below the predetermined epoch. The learning epoch for learning the artificial neural network is greater than or equal to the predetermined epoch by setting the learning rate of the artificial neural network to a predetermined value or higher. In this case, the learning rate of the aforementioned artificial neural network can be set to a predetermined value or less.

  In an alternative embodiment, means for receiving aging age learning data including the above-described physical condition information for learning, the above-described predicted life expectancy value, and aging years information for learning; performing deep learning using the above-mentioned aging age learning data And a means for receiving the physical condition information, processing the received physical condition information with the aging years prediction model, and outputting a predicted value of the aging years. Can do.

  In the embodiment as an alternative, the means for generating the aging years numerical prediction model related to health by executing the deep learning using the aging years learning data is used as the aging years prediction artificial neural network. Means for inputting; means for deriving an error by comparing the output related to the physical condition information for learning output from the artificial neural network for predicting the aging time and the aging age information for learning described above; and as a result of the comparison described above Means may be included for back-propagating the derived error to the aging prediction artificial neural network to update the weight value of the artificial neural network.

  An alternative embodiment may further include means for labeling the learning age information for each of the learning physical condition information of the age learning data described above.

  In an alternative embodiment, the means for labeling the learning age information for each of the learning physical condition information of the learning age learning data described above is that the subject included in the learning physical condition information is a dead person. In some cases, the means for labeling the learning physical condition information described above as the age information of the difference between the information related to the time included in the physical condition information for learning and the information related to the death time of the subject described above; Means for inputting the aforementioned learning physical condition information into the generated expected life expectancy prediction model when the subject included in the physical condition information is not a dead person; and the above expected life expectancy prediction model relating to the aforementioned learning physical condition information Means for labeling the above-mentioned learning physical condition information with the output as the above-mentioned aging age information can be included.

  In the embodiment as an alternative, the above-mentioned physical condition information for learning is the birth year, age, sex, height, weight, BMI, blood test values, ecological data, family calendar, drinking alcohol, exercise exercise, smoking presence and disease code Means for grouping according to a predetermined criterion, including at least one, may be further included.

  The alternative embodiment further includes means for generating the grouped group life expectancy prediction model by performing deep learning using the life expectancy learning data grouped according to the predetermined criterion. be able to.

  In an alternative embodiment, the above-described physical condition information for learning is for re-learning by processing the physical condition information for re-learning that the target person is a dead person and the target person is a non-dead person with the generated expected life expectancy prediction model. Means for generating expected life expectancy information; Means for generating re-learning data by labeling the aforementioned re-learning expected life expectancy information for each of the aforementioned re-learning physical condition information; Means for performing deep learning using the data to update the generated expected life expectancy model may be further included.

  FIG. 9 is a block diagram illustrating a module for implementing a method for predicting a numerical value related to health according to a first embodiment of the present disclosure.

  According to the first embodiment of the present disclosure, a method for predicting a numerical value related to health is a module for receiving expected life expectancy learning data including physical condition information for learning and expected life expectancy information for learning (1310); A module (1320) for generating an expected life expectancy prediction model by performing deep learning using life expectancy learning data; and receiving physical condition information, and processing the received physical condition information with the expected life expectancy prediction model This can be realized by a module (1330) for outputting a predicted value of expected life expectancy.

  FIG. 10 is a block diagram illustrating logic for embodying a method for predicting a numerical value related to health according to the first embodiment of the present disclosure.

  Based on the first embodiment of the present disclosure, a method for predicting a numerical value related to health includes logic for receiving expected life expectancy learning data including physical condition information for learning and expected life expectancy information for learning (1410); Logic for generating an expected life expectancy prediction model by executing deep learning using life expectancy learning data (1420); and physical condition information are received, and the received physical condition information is processed by the expected life expectancy prediction model. This can be realized by logic (1430) for outputting a predicted value of expected life expectancy.

  FIG. 11 is a block diagram illustrating a circuit for embodying a method for predicting a numerical value related to health according to the first embodiment of the present disclosure.

  According to the first embodiment of the present disclosure, a method for predicting a numerical value related to health is a circuit (1510) for receiving expected life expectancy learning data including physical condition information for learning and expected life expectancy information for learning; A circuit for generating an expected life expectancy prediction model by executing deep learning using life expectancy learning data (1520); and receiving physical condition information, and processing the received physical condition information with the above expected life expectancy prediction model This can be realized by a circuit (1530) for outputting a predicted value of expected life expectancy.

  Those skilled in the art will further understand that various exemplary and logical blocks, configurations, modules, circuits, means, logic, and algorithm steps in the description of the embodiments disclosed herein may be performed by electronic hardware, computer software, or a combination of both. It should be recognized that it can be implemented. To clearly illustrate the interchangeability of hardware and software, various illustrative components, blocks, configurations, means, logic, modules, circuits, and stages have been described above generally in terms of their functionality. Whether such functionality is implemented by hardware or software depends on the particular application and design constraints of the overall system. Skilled technicians can implement the functionality described in a variety of ways for each specific application, but it should not be analyzed that decisions in such implementation depart from the scope of this disclosure. Don't be.

  FIG. 12 is a simplified, general schematic diagram of an exemplary computing environment in which embodiments of the present disclosure can be implemented.

  Although this disclosure generally describes computer-executable instructions that can be executed on one or more computers, those skilled in the art will recognize that this disclosure may be combined with other program modules and / or hardware. You will appreciate that it can be implemented as a combination of hardware and software.

  Generally, program modules include routines, programs, components, data structures, and the like that perform particular tasks or implement particular abstract data types. Also, those skilled in the art will understand that the disclosed method can be applied to single-processor or multi-processor computer systems, minicomputers, mainframe computers as well as personal computers, handheld computing devices, microprocessor-based or programmable consumer electronics, and the like. It will be appreciated that other computer system configurations can be implemented, including (each of which can operate in conjunction with one or more associated devices).

  The embodiments described in this disclosure can also be practiced in distributed computing environments where certain tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules can be located in both local and remote memory storage devices.

  A computer typically includes a variety of computer-readable media. Although any computer-readable medium can be a computer-readable medium, such computer-readable media can be volatile and non-volatile media, transitory and non-transitory. Includes media, removable and non-removable media. By way of example, and not limitation, computer-readable media can include computer-readable storage media and computer-readable transmission media. Computer-readable storage media can be volatile and non-volatile media, temporary and non-volatile, implemented by any method or technique of storing information such as computer readable instructions, data structures, program modules or other data. Includes temporary media, removable and non-removable media. Computer storage media include RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, DVD (digital video disk) or other optical disk storage device, magnetic card set, magnetic tape, magnetic disk storage device or other magnetic storage device, Or includes, but is not limited to, any other medium that can be accessed by a computer and used to store necessary information.

  A computer readable transmission medium is typically a computer readable instruction word, data structure, program module or other data in a modulated data signal such as a carrier wave or other transport mechanism. Etc., including all information transmission media. The term modulated data signal means a signal that has one or more of its characteristics set or changed in such a manner as to encode information in the signal. By way of example, and not limitation, computer readable transmission media include wired media such as a wired network or direct-wired connection, and wireless media such as acoustic, RF, infrared, and other wireless media. Combinations of any of the foregoing media are also included within the scope of computer-readable transmission media.

  An exemplary environment (1600) is shown that implements various aspects of the disclosure including a computer (1602), the computer (1602) including a processing unit (1604), a system memory (1606), and a system bus (1608). including. The system bus (1608) connects system components, including but not limited to system memory (1606), to the processing unit (1604). The processing unit (1604) can be any of a variety of commercial processors. Dual processor and other multiprocessor architectures may also be utilized as the processing unit (1604).

  The system bus (1608) can be any of several types of bus structures that can be further interconnected to a local bus using any of a memory bus, peripheral bus, and various commercial bus architectures. obtain. The system memory (1606) includes read only memory (ROM) (1610) and random access memory (RAM) (1612). The basic input / output system (BIOS) is stored in a non-volatile memory (1610) such as ROM, EPROM, EEPROM, etc., but this BIOS is connected between components inside the computer (1602) when starting up. Includes basic routines that support sending information. The RAM (1612) may also include a high-speed RAM such as static RAM for caching data.

  The computer (1602) is also an internal hard disk drive (HDD) (1614) (eg, EIDE, SATA) —the internal hard disk drive (1614) is also an external hard disk in a suitable chassis (not shown). A magnetic floppy disk drive (FDD) (1616) (eg, for reading and recording data from a removable diskette (1618)), and an optical disk drive ( 1620) (for example, for reading a CD-ROM disc (1622), reading data from other high-capacity optical media such as a DVD, or recording it). The hard disk drive (1614), magnetic disk drive (1616) and optical disk drive (1620) are connected to the system bus (1608) by a hard disk drive interface (1624), a magnetic disk drive interface (1626) and an optical drive interface (1628), respectively. Can do. The interface (1624) for implementing an external drive includes at least one or both of USB (Universal Serial Bus) and IEEE 1394 interface technologies.

  These drives and associated computer readable media provide non-volatile storage of data, data structures, computer executable instructions, and the like. In the case of the computer (1602), the drive and medium correspond to storing arbitrary data in an appropriate digital format. In the above description of the computer-readable medium, HDDs, removable magnetic disks, and removable optical media such as CDs or DVDs are mentioned. Those skilled in the art will know a zip drive, a magnetic card set, a flash memory card, and a cartridge. Other types of computer readable media of this type can also be used in the exemplary operating environment, and any such media can be used to perform the disclosed methods. You will fully understand that it can contain computer executable instructions.

  A number of program modules can be stored in the drive and RAM (1612), including an operating system (1630), one or more application programs (1632), other program modules (1634), and program data (1636). . All or a portion of the operating system, applications, modules and / or data can also be cached in RAM (1612). It will be appreciated that the present disclosure can be implemented in various commercially available operating systems or combinations of operating systems.

  A user may enter commands and information into the computer (1602) through one or more wired / wireless input devices, eg, pointing devices such as a keyboard (1638) and a mouse (1640). Other input devices (not shown) may include microphones, IR remote controls, joysticks, game pads, stylus pens, touch screens, and the like. These and other input devices often connect to the processing unit (1604) through the input device interface (1642) which is connected to the system bus (1608), but the parallel port, IEEE 1394 serial port, game port, USB port, IR interface, etc. It can be connected by other interfaces of similar things.

  A monitor (1644) or other type of display device is also connected to the system bus (1608) through an interface, such as a video adapter (1646). In addition to the monitor (1644), computers typically include speakers, printers, and other peripheral output devices (not shown) of this type.

  Computer (1602) can operate in a networked environment utilizing logical connections to one or more remote computers, such as remote computer (s) (1648), via wired and / or wireless communications. . The remote computer (s) 1648 can be a workstation, computing device computer, router, personal computer, portable computer, microprocessor-based gaming device, peer device or other normal network node, In particular, it includes many or all of the components described for the computer (1602), but for the sake of simplicity, only the memory storage device (1650) is shown. The logical connections shown include a wired / wireless connection to a near field communication network (LAN) (1652) and / or a larger network, eg, a telecommunications network (WAN) (1654). Such LAN and WAN networking environments are commonplace in offices and companies, and facilitate enterprise-wide computer networks such as intranets, all of which are computer networks worldwide. For example, you can connect to the Internet.

  When used in a LAN networking environment, the computer (1602) is connected to the local network (1652) through a wired and / or wireless communication network interface or adapter (1656). The adapter (1656) can easily resolve wired or wireless communication to the LAN (1652), and this LAN (1652) can also connect a wireless access point installed on it to communicate with the wireless adapter (1656). It is included. When used in a WAN networking environment, the computer (1602) may include a modem (1658), connected to a communication computing device on the WAN (1654), or through the WAN (1654). Other means for setting communication. A modem (1658), which can be internal or external and wired or wireless, is connected to the system bus (1608) through a serial port interface (1642). In a networked environment, a plurality of program modules described in association with the computer (1602) or portions thereof can be stored in the remote memory / storage device (1650). It will be appreciated that the network connections shown are exemplary and other means of establishing a communications link between multiple computers can be used.

  Computer (1602) refers to any wireless device or object that is deployed and operating in wireless communication, eg, printer, scanner, desktop and / or portable computer, PDA (portable data assistant), communication satellite, wirelessly detectable tag Operate to communicate with any equipment or location that you want, and your phone. This includes at least Wi-Fi and Bluetooth® wireless technology. Thus, the communication can be a predefined structure as in a conventional network or simply an ad hoc communication between at least two devices.

  Wi-Fi (Wireless Fidelity) enables connection to the Internet or the like without a line. Wi-Fi is a wireless technology such as a cell phone that allows such devices, for example, a computer to send and receive data both indoors and outdoors, i.e. anywhere within the coverage area of a base station. Wi-Fi networks use wireless technology called IEEE 802.11 (a, b, g, etc.) to provide secure, reliable and high speed wireless communication. Wi-Fi can be used to connect computers to each other, the Internet, and wired networks (using IEEE 802.3 or Ethernet). Wi-Fi networks operate in unlicensed 2.4 and 5 GHz radio bands, for example, at products with 11 Mbps (802.11a) or 54 Mbps (802.11b) data rates, or in products that include both bands (dual bands) can do.

  Those of ordinary skill in the art of the present disclosure will appreciate that information and signals can be represented utilizing any of a variety of different technologies and techniques. For example, data, instructions, instructions, information, signals, bits, symbols and chips that can be referenced in the above description are represented by voltage, current, electromagnetic wave, magnetic field or particle, optical field or particle, or any combination thereof. Can.

  Those having ordinary skill in the art of this disclosure will appreciate that various exemplary logic blocks, modules, processors, means, circuits, and algorithm steps in the description of the embodiments disclosed herein are electronic hardware ( It will be appreciated that various forms of program or design code (referred to herein as “software” or a combination of all of these may be implemented for convenience). To clearly illustrate such interoperability of hardware and software, various illustrative components, blocks, modules, circuits, and stages have been generally described above for these functions. Whether such functionality is implemented by hardware or software depends on the particular application and design constraints in the overall system. Those having ordinary skill in the art of the present disclosure may implement the functions described in various ways for each specific application, but such implementation decisions are beyond the scope of this disclosure. It should not be parsed to do.

  The various embodiments shown herein can be implemented in the form of methods, apparatus, or articles using standard programming and / or engineering techniques. As used herein, the term “article of manufacture” includes a computer program, carrier, or media accessible from any computer readable storage device. For example, computer readable storage media include magnetic storage devices (eg, hard disks, floppy disks, magnetic strips, etc.), optical disks (eg, CDs, DVDs, etc.), smart cards, and flash memory devices (eg, EEPROMs, cards). , Sticks, key drives, etc.), but are not limited to these. The various storage media shown herein also include one or more devices and / or other machine-readable media for storing information.

  It should be understood that the specific order or hierarchy of steps in the illustrated process is an example of an exemplary approach. It should be understood that a specific order or hierarchy of steps in the process can be rearranged within the scope of this disclosure based on design priority. The accompanying method claims are sample order and provide elements of various stages, but are not meant to be limited to the specific order or hierarchy presented.

  The description of the illustrated embodiments is provided to enable any person skilled in the art to use or practice the present disclosure. Various modifications to such embodiments will be apparent to those of ordinary skill in the art of this disclosure, and the general principles defined herein may be used in other ways without departing from the scope of this disclosure. It can be applied to the embodiment. Accordingly, the present disclosure is not limited to the embodiments shown herein but should be analyzed in the broadest sense consistent with the principles and novel features shown herein.

Claims (16)

In a method for predicting health-related figures,
Receiving expected life expectancy learning data including physical condition information for learning and expected life expectancy information for learning;
Executing deep learning using the expected life expectancy learning data to generate an expected life expectancy model; and receiving physical condition information, processing the received physical condition information with the expected life expectancy model, and expecting life expectancy Outputting a predicted value of;
including,
A method to predict health figures. In claim 1,
Comparing the output predicted life expectancy value and the statistical life expectancy information corresponding to the physical condition information to calculate at least one of a predicted age value and a predicted health age value;
Further including
A method to predict health figures. In claim 1,
The learning physical condition information and the physical condition information are:
Name, personal ID, year of birth, age, gender, personal information, health checkup date, height, weight, BMI, blood test data, ecological data, family calendar, alcohol consumption, exercise status, smoking status, treatment date, medical treatment Including at least one of institutional information, medical breakdown, life / death, date of death, and disease code,
A method to predict health figures. In claim 1,
Labeling the expected life expectancy information for each of the physical condition information for learning of the expected life expectancy learning data;
Further including
A method to predict health figures. In claim 4,
Labeling the expected life expectancy information for each of the physical condition information for learning of the expected life expectancy learning data,
When the subject included in the physical condition information for learning is a dead person, the difference between the information related to the time included in the physical condition information for learning and the information related to the death time of the target person is the expected life expectancy information, Labeling the learning physical condition information; and, if the subject included in the learning physical condition information is not a dead person, the expected life expectancy information based on the personal information of the subject as the expected life expectancy information, Labeling the learning physical condition information;
including,
A method to predict health figures. The step of performing deep learning using the expected life expectancy learning data and generating the expected life expectancy model in claim 1,
Inputting the learning physical condition information into an artificial neural network;
Deriving an error by comparing the output related to the physical condition information for learning output from the artificial neural network with the expected life expectancy for learning; and back-propagating the error derived as a result of the comparison to the artificial neural network Updating the weight value of the artificial neural network;
including,
A method to predict health figures. In claim 1,
The step of performing deep learning using the learning data and generating the expected life expectancy model includes:
If the learning epoch for learning the artificial neural network is less than or equal to a predetermined epoch, a learning rate of the artificial neural network is set to a predetermined numerical value or more, and the artificial If the learning epoch for learning the neural network is greater than or equal to a predetermined epoch, the learning rate of the artificial neural network is set below a predetermined numerical value,
A method to predict health figures. In claim 1,
Receiving age learning data including the learning physical condition information, the predicted life expectancy value, and learning age information;
Performing deep learning using the aging years learning data to generate a aging years prediction model; and receiving the physical condition information and processing the received physical condition information with the aging years prediction model Outputting the predicted number of years;
Further including
A method to predict health figures. The step of performing deep learning using the aging years learning data and generating a numerical model for predicting aging years related to health in claim 8,
Inputting the physical condition information for learning into an artificial aging prediction neural network;
A step of deriving an error by comparing the learning age information output from the learning physical condition information output from the learning age prediction artificial neural network; and the error derived as a result of the comparison Backpropagating to the predictive artificial neural network to update the weight value of the artificial neural network;
including,
A method to predict health figures. In claim 9,
Labeling the learning age information on each of the learning physical condition information of the age learning data;
Further including
A method to predict health figures. In claim 10,
Labeling the learning age information on each of the learning physical condition information of the age-related learning data,
When the subject included in the physical condition information for learning is a dead person, the difference between the information related to the time included in the physical condition information for learning and the information related to the death time of the target person is set as age information, and the learning Labeling physical condition information;
If the subject included in the learning physical condition information is not a dead person, inputting the learning physical condition information into the generated expected life expectancy model; and the expected life expectancy prediction model related to the learning physical condition information; Labeling the learning physical condition information with the output as the age information;
including,
A method to predict health figures. In claim 1,
The learning physical condition information includes at least one of the birth year, age, sex, height, weight, BMI, blood test values, ecological measurement data, family calendar, alcohol consumption, exercise activity, smoking activity, and disease code in advance. Grouping according to established criteria;
Further including
A method to predict health figures. In claim 12,
Performing deep learning using expected life expectancy learning data grouped according to the predetermined criteria to generate the grouped group expected life expectancy prediction model;
Further including
A method to predict health figures. In claim 1,
The learning physical condition information is a step of generating re-learning expected life expectancy information by processing the re-learning physical condition information in which the target person is a dead person and the target person is a non-dead person with the generated expected life expectancy model ;
Labeling the expected life expectancy information for re-learning on each of the physical information for re-learning to generate re-learning data;
Performing deep learning using the expected life expectancy learning data and the re-learning data to update the generated expected life expectancy prediction model;
Further including
A method to predict health figures. A device for predicting health related values,
A receiving unit for receiving life expectancy learning data including physical condition information for learning and life expectancy information for learning;
A learning unit that performs deep learning using the expected life expectancy learning data and generates an expected life expectancy prediction model; and receives physical condition information, and processes the received physical condition information with the expected life expectancy prediction model to predict the expected life expectancy Health-related prediction model that outputs
including,
A device that predicts values related to health. A computer program stored on a computer readable storage medium containing encoded instructions, wherein the computer program is executed by one or more processors of a computer system, the one or more processors are healthy Try to perform the following steps to make the numbers predictable, which steps are:
Receiving expected life expectancy learning data including physical condition information for learning and expected life expectancy information for learning;
Executing deep learning using the expected life expectancy learning data to generate an expected life expectancy model; and receiving physical condition information, processing the received physical condition information with the expected life expectancy model, and expecting life expectancy Outputting a predicted value of;
including,
A computer program stored on a computer-readable storage medium.