JP2018124702A - Etiological analysis device and disease prediction device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To effectively prevent diseases by checking eating habits.SOLUTION: In a disease prevention system, an etiological analysis device acquires ingestion information indicative of food items ingested by a user and refers to morbidity information indicative of diseases which the use has had, to calculate levels of correlation between the diseases and the food items as levels of food contribution. A disease prediction device refers to the ingestion information and etiological information defining the levels of food contribution to predict a disease which the user may have with high possibility. The ingestion information is acquired from a questionnaire and meal images. Further, the disease prediction device suggests food items for reducing the risk of the disease, as recommended food items.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a technique for searching etiology by checking dietary habits.

  It is said that the annual medical cost per Japanese is about 310,000 yen, especially about 930,000 yen for those over 75 years old (announced by the Ministry of Health, Labor and Welfare in 2014). With the aging of society and the advancement of medical technology, the trend of increasing medical expenses is expected to continue.

  Disease prevention is important in order to curb medical costs and extend healthy life expectancy. Patent Document 1 discloses a system that provides health advice to a user (examinee) based on the result of a health checkup.

JP 2010-277574 A

  However, when an abnormality is determined in a health checkup, it is often already sick or one step before that. Medical examinations are usually performed only once a year. Due to time constraints, it is difficult to increase the frequency of medical checkups. In addition, increasing the number of medical checkups may further increase medical costs.

  The present invention has been completed based on the above-mentioned problem recognition, and its main object is to provide a technique for effectively preventing illness by checking dietary habits.

  A pathogenesis analysis apparatus according to an aspect of the present invention includes a feeding information acquisition unit that acquires feeding information indicating a food ingested by a user, a morbidity information acquisition unit that acquires morbidity information indicating a disease affected by the user, The contribution calculation unit that calculates the degree of correlation between the disease and the food as the food contribution degree by referring to the feeding information and the morbidity information of each of the users, and the food contribution degree of each of the disease and the multiple types of foods to the disease A pathogenesis information output unit that outputs the pathogenesis information.

  A disease prediction apparatus according to an aspect of the present invention includes a feeding information acquisition unit that acquires feeding information indicating food taken by a user, and etiology information that defines a degree of food contribution indicating a degree of correlation between the disease and the food. A disease prediction unit that refers to the feeding information and predicts a disease that is likely to affect the user.

  The disease prediction apparatus according to another aspect of the present invention includes a feeding information acquisition unit that acquires feeding information indicating food taken by a user, and etiology information that defines a degree of nutritional contribution indicating a magnitude of correlation between the disease and a nutrient. And a disease predicting unit that predicts a disease that is likely to cause the user based on nutrients contained in the food ingested by the user.

A disease prediction system according to an aspect of the present invention includes an etiology analyzer and a disease predictor.
The etiology analyzer includes a feeding information acquisition unit that acquires feeding information indicating food taken by the first user, a disease information acquisition unit that acquires disease information indicating a disease affected by the first user, and a plurality of disease information acquisition units. The contribution calculation unit that calculates the degree of correlation between the disease and the food as the food contribution degree by referring to the feeding information and the morbidity information of each of the users, and the food contribution degree of each of the disease and the multiple types of foods to the disease A pathogenesis information output unit that outputs the pathogenesis information as pathogenesis information.
The disease prediction apparatus acquires a second user's feeding information based on a question transmission unit that transmits questionnaire information for selecting and inputting food ingested by the second user, and the second user's answer to the questionnaire information The feeding information acquisition unit, the etiology information and the second user's feeding information, a disease prediction unit that predicts a disease that is likely to cause the user, and the questionnaire information And a questionnaire setting section for updating questions.

  According to this invention, it becomes easy to prevent a disease effectively by checking dietary habits.

It is a hardware block diagram of a disease prediction system. It is a functional block diagram of an etiology analyzer. It is a data structure figure of feeding information. It is a data structure figure of disease information. It is a conceptual diagram of an etiology model. It is a data structure figure of etiology information. It is a data structure figure of nutrition information. It is a functional block diagram of a disease prediction device. It is a data structure figure of disease risk information. It is a flowchart which shows the process of a disease prediction. It is a flowchart which shows the meal guidance information generation process in S14 of FIG. It is a screen figure of a disease prediction screen. It is a screen figure of a meal guidance screen. It is a screen figure of an inspection proposal screen.

FIG. 1 is a hardware configuration diagram of the disease prediction system 100.
The disease prediction system 100 includes an etiology analyzer 102 and a disease predictor 104. The etiology analyzer 102 identifies foods that affect the disease. As will be described in detail later, the etiology analyzer 102 accumulates meal histories of a large number of users (hereinafter referred to as “feeding information”) and information on the diseases affected by the users (hereinafter referred to as “morbidity information”). Then, the causal relationship between the meal and the disease is modeled by a neural network (hereinafter referred to as “pathogenesis model”). The disease prediction device 104 predicts a disease that a user may suffer from based on the etiology model, and provides advice for disease prevention.

  The etiology analyzer 102 and the disease predictor 104 are connected to each other via a communication line. The disease prediction system 100 is connected to the hospital 108 and a plurality of client terminals 106a, 106b,... 106n (hereinafter collectively referred to as “client terminal 106” when collectively referred to or not particularly distinguished) via the Internet 110. The The client terminal 106 in this embodiment may be a mobile terminal such as a smartphone, or a general-purpose computer such as a laptop PC. The client terminal 106 and the Internet 110 are wirelessly connected, but may be wired. A user ID is given in advance to the user of the client terminal 106.

  In the following, we will explain the etiology analysis to search for foods that affect the disease, and then explain the disease prediction based on the etiology model.

[Etiology analysis]
FIG. 2 is a functional block diagram of the etiology analyzer 102.
As described above, the disease prediction system 100 includes the etiology analyzer 102 and the disease predictor 104. Each component of the etiology analyzing apparatus 102 and the disease predicting apparatus 104 includes a computing unit such as a CPU (Central Processing Unit) and various coprocessors, a storage device such as a memory and a storage, and a hardware including a wired or wireless communication line connecting them. Hardware and software that stores processing instructions to a computing unit. The computer program may be constituted by a device driver, an operating system, various application programs located in an upper layer thereof, and a library that provides a common function to these programs. Each block described below is not a hardware unit configuration but a functional unit block.
The etiology analyzer 102 and the disease prediction device 104 may include a web server, and the client terminal 106 may include a portable communication terminal and a web browser installed therein. .

As a user of the etiology analyzer 102, a researcher who searches for the etiology based on eating information and morbidity information is assumed.
The etiology analyzer 102 includes a user interface processing unit 120, a communication unit 124, a data processing unit 122, and a data storage unit 126.
The user interface processing unit 120 receives operations from the user via an input interface such as a touch panel, a mouse, and a keyboard, and is also responsible for processing related to the user interface such as image display and audio output. The communication unit 124 is in charge of communication processing with the disease prediction device 104, the client terminal 106, and the hospital 108 via the Internet 110. The data storage unit 126 stores various data. The data processing unit 122 executes various processes based on the data acquired by the user interface processing unit 120 and the communication unit 124 and the data stored in the data storage unit 126. The data processing unit 122 also functions as an interface for the user interface processing unit 120, the communication unit 124, and the data storage unit 126.

The communication unit 124 includes a feeding information acquisition unit 136 and a disease information acquisition unit 138.
The feeding information acquisition unit 136 receives feeding information from the client terminal 106 or the hospital 108. The feeding information is obtained from the user's answer to the selective questionnaire. In the questionnaire, a question such as “Please select the food you ate on the morning of January 23, 2017 from the following” is set, and the user can choose from multiple choices such as “egg, nori, rice, bread” Feeding information is provided to the etiology analyzer 102 by selecting the selected food.

  The user may take a meal image with the client terminal 106 during a meal, and transmit the meal image from the client terminal 106 to the etiology analyzer 102. The eating information acquisition unit 136 may identify the food ingested by the user by receiving the meal image and subjecting the meal image to image processing. When a user has a meal at a restaurant, the menu selected by the restaurant or the user may be transmitted to the etiology analyzer 102. The feeding information acquisition unit 136 may receive the menu information and specify the food associated with the menu information as the food that the user has consumed.

  The disease information acquisition unit 138 acquires disease information from the client terminal 106 or the hospital 108. The hospital 108 transmits the diagnosis result of the health check to the etiology analyzer 102. The disease information acquisition unit 138 acquires this diagnosis result as disease information. The hospital 108 may diagnose a patient and send the diagnosis result to the etiology analyzer 102 as morbidity information. Further, the user may notify the etiology analyzer 102 of the disease that has been affected by self-report. Since the disease information is deeply related to privacy, it is necessary to provide it to the etiology analyzer 102 after obtaining permission from the user.

The data processing unit 122 includes a contribution calculation unit 134.
The contribution degree calculation unit 134 calculates the likelihood of causing various food diseases (hereinafter referred to as “food contribution degree”). The contribution calculation unit 134 also calculates the likelihood of causing various nutrients (hereinafter referred to as “nutrition contribution”). For example, based on the eating information of a user who has suffered a stroke, information is extracted that if the egg is ingested, it is likely to cause a stroke. At this time, the magnitude of the influence of the egg on the stroke is indexed as the food contribution. A method for calculating the food contribution and the nutrition contribution will be described later.

  The user interface processing unit 120 includes an input unit 128 that receives input from the user, and an output unit 130 that outputs various types of information such as images and sounds to the user. The output unit 130 further includes an etiology information output unit 132. The etiology information output unit 132 displays a list of foods or nutrients that affect the disease (hereinafter referred to as “pathogenesis information”) based on the various contributions calculated by the contribution calculation unit 134. The etiology information output unit 132 may transmit the etiology information to the outside via the Internet 110.

FIG. 3 is a data structure diagram of the feeding information 140.
The feeding information 140 is stored in the data storage unit 126 of the etiology analyzer 102. The disease prediction device 104 can also acquire similar feeding information 140 (described later). The feeding information 140 is defined for each user (subject). FIG. 3 shows a history of meals of the user R. The user R is eating “egg” at the lunch on January 16, 2017 and the lunch on the 17th. He also eats “cup ramen” for dinner on the 16th and dinner on the 17th. On the 16th and 17th, we have “tea” for breakfast, lunch and dinner. The etiology analyzer 102 continuously collects a large amount of feeding information 140 from a large number of users.

  The output unit 130 of the etiology analyzer 102 displays a questionnaire screen for the user (not shown). The questionnaire screen may be displayed on the screen of the etiology analyzer 102 or may be displayed on the client terminal 106. The user logs in to the etiology analyzer 102 using the user ID, and provides the feeding information 140 to the etiology analyzer 102 by answering a questionnaire every time they eat. As described above, instead of answering the questionnaire, the user may transmit a meal image to the etiology analyzer 102.

  In the questionnaire, food names such as “egg” and “cup ramen” may be selected, or a meal menu such as “oyster fried”, “hamburger”, and “seafood bowl” may be selected. For example, hamburger includes foods such as “beef”, “flour”, and “onion”. The meal menu and food are associated in advance as “menu information”. When the user selects a meal menu through a questionnaire, the feeding information acquisition unit 136 refers to the menu information, identifies the food eaten by the user, and records it as the feeding information 140.

FIG. 4 is a data structure diagram of the disease information 150.
The disease information 150 is stored in the data storage unit 126. The disease information 150 indicates user attributes such as the age and sex of each user, and their medical history. The disease information 150 is provided from the hospital 108. The user may self-report the medical history from the client terminal 106 to the etiology analyzer 102. The disease information acquisition unit 138 acquires the disease information 150 and registers it in the data storage unit 126.

  Referring to the morbidity information 150 in FIG. 4, the user with user ID = P001 (hereinafter referred to as “user (P001)”) is a 24-year-old woman and has no medical history such as stroke or stomach cancer. The user (P003) is a 44-year-old male and suffers from a stroke when he is 42 years old. The user (P005) is a 74-year-old woman who has stomach cancer at the age of 58.

  The etiology analyzer 102 matches the feeding information 140 and the morbidity information 150 to find out what kind of illness the person taking the food is likely to get, what kind of illness the person taking the food is less likely to get. To analyze. Next, the analysis method will be described.

FIG. 5 is a conceptual diagram of the etiology model 160.
The etiology model 160 in the present embodiment is formed by a neural network. Input layers X1, X2,... Xr indicate food. For example, X1 corresponds to “egg” and X2 corresponds to “cup ramen”. Y1 and Y2 in the output layer indicate the presence or absence of a certain disease. For example, Y1 indicates “affected by stroke” and Y2 indicates “not affected by stroke”. The number of intermediate layers U1, U2... Us and the number of hierarchies are arbitrary. In the present embodiment, the description will be made assuming that the contribution calculation unit 134 generates the etiology model 160 for each disease such as stroke, stomach cancer, and heart disease.

  For example, a user suffers from a stroke. First, the user's feeding information 140 is set in the input layer (X1, X2,... Xr), and “1” is set in the output layer Y1 and “0” is set in Y2. As is known, the input layer and the intermediate layer, the intermediate layer and the output layer are connected by an arbitrary activation function, for example, a logistic sigmoid function. When this user (stroked person) has consumed a large amount of egg (X1) in the past, but has not consumed a large amount of sardine (X5), the connection between the egg (X1) and the output node Y1 (weight coefficient) ) Is strengthened, and the connection between the sardine (X5) and the output node Y1 is weakened. Further, the connection between the egg (X1) and the output node Y2 is weakened, and the connection between the sardine (X5) and the output node Y2 is strengthened.

  The input value of the egg (X1) is the number of eggs eaten or consumed, such as the number and number of eggs eaten in the past meals, the frequency of eating eggs during meals (how many times the egg is eaten at a rate of 1 meal) What is necessary is just to define arbitrarily as a meal rate.

  Learning based on the feeding information 140 and the morbidity information 150 models what kind of food affects what kind of disease in the etiology model 160. A food contribution degree is calculated based on the learned etiology model 160.

If the strength (weighting coefficient) of the connection between the output layer Y1 and the intermediate layer U1 is c11, and the strength of the connection between the output layer Y1 and the intermediate layer U2 is c21,
y1 = c11 · u1 + c21 · u2 + ...
Can be expressed as y1 is an input value of the output layer Y1, and u1 and u2 are output values of the intermediate layers U1 and U2.

If the strength of the connection between the intermediate layer U1 and the input layer X1 is a11 and the strength of the connection between the intermediate layer U1 and the input layer X2 is a21,
u1 = a11 · x1 + a21 · x2 +...
Can be expressed as x1 and x2 are output values of the input layers X1 and X2. When the contribution degree calculation unit 134 calculates the food contribution degree, for convenience, the input value and the output value of the intermediate layer may be regarded as the same (identity mapping).

Similarly, if the strength of the connection between the intermediate layer U2 and the input layer X1 is a12 and the strength of the connection between the intermediate layer U2 and the input layer X2 is a22,
u2 = a12 · x1 + a22 · x2 + ...
Can be expressed as u2 is an input value of the intermediate layer U2.

With the above,
y1 = (a11 · c11 + a12 · c21 +...) · x1 +.
Can be expressed as At this time, the data processing unit 122 calculates the magnitude of the influence (food contribution degree) of the egg (x1) on the stroke (y1) as “a11 · c11 + a12 · c21 +.

  As described above, the contribution calculation unit 134 repeats the “supervised learning” by using a large amount of feeding information 140 and morbidity information 150, so that the etiology model 160 is appropriately selected as a food for the disease of the food in parallel with the refinement. Calculate the contribution.

  In the present embodiment, a plurality of etiology models 160 are generated according to the age group and according to the disease. For example, the etiology model 160 for determining the presence / absence of stroke for users 21 to 30 years old and the etiology model 160 for determining the presence / absence of strokes for users 31 to 40 years old are different. Is generated. Specifically, y1 = 1 and y2 = 0 are set for a user who has developed a stroke between the ages of 21 and 30, and y1 = 0 and y2 = 1 are set for a user who has not developed a stroke. After setting, the feeding information 140 is set as an input value to adjust the coefficient such as c11. By such a method, the etiology model 160 that predicts the risk of developing a stroke between the ages of 21 and 30 is formed.

  Instead of generating the etiology model 160 for each disease such as stroke or stomach cancer, a single etiology model 160 may be used to determine a plurality of diseases. For example, by setting the output layer Y1 to determine the presence / absence of a stroke and the output layer Y2 to determine the presence / absence of gastric cancer, it is possible to perform a plurality of disease determinations using one etiology model 160.

FIG. 6 is a data structure diagram of the etiology information 170.
The etiology information 170 is stored in the data storage unit 126. The etiology information 170 is defined for each disease. FIG. 6 shows food contributions of various foods to “stroke”. According to the etiology information 170 of FIG. 6, there is a strong positive correlation between “cup ramen” and “stroke”. The food contribution level of “8.5” quantitatively indicates that “cup ramen” is likely to cause “stroke”. On the other hand, there is a strong negative correlation between “tomato” and “stroke”. The food contribution “−7.7” quantitatively indicates that “tomato” suppresses “stroke”. From the etiology information 170 of FIG. 6, it can be found that in order not to suffer from a stroke, it is necessary to refrain from cup ramen and actively ingest tomatoes.

  Hereinafter, foods that easily cause illness such as “cup ramen” for “stroke” are called “disease-promoting foods”, and foods that suppress illness such as “tomatoes” are called “disease-suppressing foods”. The contribution calculation unit 134 sets a “food rank” for each disease according to the magnitude of the food contribution. In the present embodiment, seven levels of food ranks S, A, B, C, D, E, and F are set. S rank is the most effective disease-suppressing food for disease prevention, and F rank indicates a disease promoting food that is most likely to cause the disease. The contribution degree calculation unit 134 sets the food rank such that food having a food contribution degree equal to or greater than the first threshold is F rank, and food that is greater than or equal to the second threshold and less than the first threshold is E rank.

FIG. 7 is a data structure diagram of the nutrition information 180.
The nutrition information 180 is stored in the data storage unit 126. The nutrition information 180 is defined for each food. FIG. 7 shows the amount of nutrients (components) contained in “cup ramen”. The data of the nutrition information 180 is set by the designer based on nutrition. According to the nutrition information 180 of FIG. 7, “cup ramen” contains only 25.2 micrograms of vitamin A and 6.21 grams of salt. The contribution calculation unit 134 ranks the content of nutrients for each food (hereinafter referred to as “nutrition rank”). In the present embodiment, seven levels of nutrition ranks A1, A2, A3, B1, B2, C1, and C2 are set. A1 rank nutrients are contained in a large amount in the food, and C2 rank nutrients are hardly contained in the food. The nutritional rank indicates how much the target nutrient is contained in the food.

The nutrition rank of nutrient B in a food A is the standard deviation of the content of nutrient B in food A when the content of nutrient B in all registered foods (eg, mass percent concentration) is used as the population. Is indexed. According to FIG. 7, “cup ramen” contains much more “salt” compared to other foods and hardly contains “vitamin A” compared to other foods. The contribution degree calculation unit 134 sets the nutrition rank such that food having a standard deviation equal to or greater than the third threshold is A1 rank, and food whose standard deviation is equal to or greater than the fourth threshold and less than the third threshold is A2 rank.
The nutrition rank may also be set in advance in the nutrition information 180 based on existing knowledge by the designer.

  Disease-promoting foods that have a high food contribution to stroke, for example, all of the top three food-promoting foods that have a high nutritional rank (eg, A2 or higher) in “salt” (disease promotion) If the food contains a lot of salt), it can be estimated that salt is likely to cause a stroke. On the other hand, when a high nutritional rank (e.g., A1 or higher) is given to “dietary fiber” in disease-suppressing foods with a small food contribution to stroke, for example, all of the disease-suppressing foods with food contributions up to the bottom three It can be estimated that dietary fiber tends to suppress stroke (when dietary fiber contains a lot of dietary fiber).

  Hereinafter, nutrients that are likely to cause illness such as “salt” for “stroke” are referred to as “disease promoting nutrients”, and nutrients that suppress illness such as “dietary fiber” are referred to as “disease inhibiting nutrients”.

  The contribution calculation unit 134 calculates the nutrition contribution based on the food contribution. The coefficients 0.1, 0.07, 0.05,... Are associated with the nutrition ranks A1, A2, A3,. Since the food contribution degree of “cup ramen” to the stroke is 8.5, the contribution degree calculation unit 134 has a salt contribution degree (based on the cup ramen) of 8.5 × 0.1 = 0.85 to the stroke. calculate. When the nutritional rank of the salt contained in the egg is A2, the food contribution of “egg” to the stroke is 2.8, so the contribution calculating unit 134 determines the contribution of salt (based on the egg) to the stroke. 2.8 × 0.07 = 0.196 is calculated. Similarly, the degree of contribution of the salt through each food to the stroke is calculated, and the total value is calculated as the “nutrient contribution to the salt” for the stroke.

  The above is an example of a method for calculating the degree of nutrition contribution. For example, the nutrient contribution may be calculated by multiplying the standard deviation of the content of nutrient B in food A described above by the food contribution of food A.

  According to the above processing method, it is possible to specify the disease-promoting nutrients contained in the disease-promoting food and the disease-suppressing nutrients contained in the disease-suppressing food. In general, it is better to refrain from salt in order to prevent stroke, but such advice is often based on doctors' rules of thumb. The etiology analyzer 102 can identify disease-promoting foods and disease-suppressing foods on a fact base from the enormous feeding information 140 and morbidity information 150, and can further identify nutrients that have an influence on the disease.

  The etiology information output unit 132 lists food contributions for each combination of disease and food and outputs the list as the etiology information 170. In addition, the etiology information output unit 132 can output the etiology information 170 including the nutrition contribution for each combination of disease and nutrient.

[Disease prediction]
FIG. 8 is a functional block diagram of the disease prediction apparatus 104.
The disease prediction device 104 predicts a disease (hereinafter referred to as “attention-related disease”) that is likely to cause the user in the future based on the user's feeding information 140 and the etiology model 160. A user who wants to receive the disease prediction service of the disease prediction device 104 needs to provide the feeding information 140 to the disease prediction device 104 after registering as a user in the disease prediction device 104. When the feeding information 140 of the user is stored in the etiology analyzer 102, the feeding information 140 of the user may be provided from the etiology analyzer 102 to the disease prediction device 104.

The disease prediction device 104 includes a user interface processing unit 190, a communication unit 192, a data processing unit 196, and a data storage unit 198.
The user interface processing unit 190 receives operations from the user through an input interface such as a touch panel, a mouse, and a keyboard, and is also in charge of processing related to the user interface such as image display and audio output. The communication unit 192 is in charge of communication processing with the etiology analyzer 102, the client terminal 106, and the hospital 108 via the Internet 110. The data storage unit 198 stores various data. The data processing unit 196 executes various processes based on the data acquired by the user interface processing unit 190 and the communication unit 192 and the data stored in the data storage unit 198. The data processing unit 196 also functions as an interface for the user interface processing unit 190, the communication unit 192, and the data storage unit 198.

The communication unit 192 includes a feeding information acquisition unit 204 and a question transmission unit 206.
The feeding information acquisition unit 204 receives the feeding information 140 from the client terminal 106 or the hospital 108. The feeding information 140 may be obtained from a user's answer to a selection type questionnaire. The feeding information acquisition unit 204 may specify a food included in the menu by image processing from a meal image captured by the user. In addition, when the user has a meal at the restaurant, the feeding information 140 may be acquired from the menu information selected by the user. The question transmission unit 206 transmits a questionnaire to the client terminal 106. The user provides the feeding information 140 to the disease prediction device 104 by answering the questionnaire.
The feeding information acquisition unit 204 may receive the feeding information 140 from the etiology analyzer 102. Similarly, the feeding information acquisition unit 136 of the etiology analyzer 102 may receive the feeding information 140 acquired by the disease prediction device 104.

The data processing unit 196 includes a disease prediction unit 208, a meal guidance unit 210, a life guidance unit 212, a questionnaire setting unit 213, and a test proposal unit 214.
The questionnaire setting unit 213 generates questionnaire information for acquiring the feeding information 140. The disease prediction unit 208 predicts a disease that may be affected in the future based on the feeding information 140 and the etiology model 160 (described later). The meal instruction unit 210 provides the user with meal instruction information for reducing the possibility of illness. The life guidance unit 212 provides life guidance information for reducing the possibility of morbidity. The examination proposing unit 214 proposes an optional examination in the health examination. These will also be described later.

  The user interface processing unit 190 includes an input unit 200 that receives input from the user, and an output unit 202 that outputs various types of information such as images and sounds to the user.

FIG. 9 is a data structure diagram of the disease risk information 220.
The disease risk information 220 is stored in the data storage unit 198. The disease risk information 220 is defined for each disease. FIG. 9 shows the high risk of suffering from “stroke”. For example, a user (P001: 24 year old female) has a 0.0 percent chance of having a stroke by 30 years old, but 1.2 percent from 31 to 41 years old and 8 from 41 to 50 years old. .4% risk. The user (P003: 44-year-old male) has experienced a stroke by the age of 41 to 50, and has a recurrence risk of 36.5% from the age of 51 to 60.

  The disease prediction unit 208 calculates the risk of morbidity based on the etiology model 160 shown in FIG. Specifically, feeding information of the user (P001: 24-year-old woman) is input to the input layer of the etiology model 160 generated based on the morbidity information 150 and the feeding information 140 for people aged 21 to 30 years old. By setting 140, the possibility that this user will suffer from a stroke by 21 to 30 years old is obtained. In the present embodiment, the risk is calculated based on the magnitude of y1 (stroke) in the etiology model 160 of FIG. y1 = 1.0 corresponds to 100 percent and y1 = 0.0 corresponds to 0 percent.

  Similarly, the disease predicting unit 208 adds the feeding information 140 of the user (P001: 24-year-old woman) to the etiology model 160 generated based on the morbidity information 150 and the feeding information 140 of people 31 to 40 years old. , The possibility that this user will suffer from a stroke by 31 to 40 years old is also obtained. With the above processing method, the risk of a user (P001: 24-year-old woman) suffering from a stroke in the future is predicted. The same applies to other diseases.

FIG. 10 is a flowchart showing a disease prediction process.
First, the feeding information acquisition unit 204 receives the feeding information 140 from the client terminal 106 of the user who wants to receive the disease prediction service (S10). The disease prediction unit 208 calculates morbidity risk for various diseases based on the etiology model 160 (S12). For example, in the case of a user in his twenties, a future morbidity risk is calculated based on the etiology model 160 corresponding to each age group of 21-30 years old, 31-40 years old,. To record.

  Next, based on the disease risk information 220, the meal guidance unit 210 generates meal guidance information (S14). The dietary guidance information is information that suggests a disease-suppressing food effective in suppressing the risk of suffering from a disease requiring attention and advises to refrain from a disease-promoting food. Details of the processing of S14 will be described in detail with reference to FIG. Details of the meal guidance information will be described in detail with reference to FIG.

  The life guidance unit 212 generates life guidance information (S16). The life guidance information is information indicating advice on life that is effective in suppressing the risk of the user suffering from the disease requiring attention. The life guidance information will also be described later with reference to FIG.

  The inspection proposal unit 214 generates inspection proposal information (S18). The examination proposal information is information for proposing an examination that is effective for finding a disease requiring attention. The inspection proposal information will be described later with reference to FIG. Finally, the disease prediction unit 208 and the like display the diagnosis result by transmitting the disease prediction result, meal guidance information, and the like to the client terminal 106 via the communication unit 192 (S20).

FIG. 11 is a flowchart showing the meal guidance information generation process in S14 of FIG.
The meal guidance unit 210 first extracts “recommended food” (S30). Recommended foods are selected from disease-suppressing foods for diseases requiring attention. For example, when the stroke is a disease requiring attention, a disease-suppressing food having a food rank of a predetermined rank or less, such as tomato and sardine, is selected as a recommended food. Further, recommended foods may be additionally selected based on nutrients. For example, when the nutritional contribution of dietary fiber to stroke is low (suppressive), a food containing a large amount of dietary fiber (disease-suppressing nutrient) may be selected as the recommended food.

  Next, the meal guidance unit 210 extracts “food requiring attention” (S32). The food requiring attention is selected from disease-promoting foods for the disease requiring attention. For example, when a stroke is a disease requiring attention, a disease promoting food having a food rank of a predetermined rank or higher, such as cup ramen or takuan, is selected as a food requiring attention. In addition, foods requiring attention may be additionally selected based on nutrients. For example, when a carbohydrate has a high nutritional contribution to stroke (strong promotion of stroke), a food containing a large amount of carbohydrate (disease-promoting nutrient) may be selected as a food requiring attention.

  Correspondence relations may be determined in advance for a plurality of foods such as natto and yudofu, hijiki and seaweed. In this case, when the meal guidance unit 210 extracts natto as recommended food or cautionary food, the tofu curd previously associated with natto is also extracted as recommended food or cautionary food.

  Next, the meal guidance unit 210 performs a user allergy check (S34). The user's allergy information may be self-reported by the user, or the disease prediction device 104 may share allergy information with the hospital 108. The meal guidance unit 210 excludes “soba” from the recommended food list for the soba allergy user. Such a treatment method makes it possible to propose a dietary habit to prevent the onset of illness while avoiding the risk of allergies. Since allergies are often alleviated as they grow, it is desirable for users to regularly update allergy information.

FIG. 12 is a screen diagram of the disease prediction screen 230.
The disease prediction screen 230 is generated by the disease prediction unit 208 and transmitted to the client terminal 106. The disease prediction screen 230 is provided in the form of a web page. The same applies to the meal guidance screen described below. FIG. 12 shows a disease prediction screen 230 provided for a certain user P (Nakano Kyoko (a pseudonym) / 28 years old female).

  As described with reference to FIG. 9, the risk of suffering from various diseases is predicted based on the user's feeding information 140 and the etiology model 160. The user P has various caution-related diseases. Among them, stroke, diabetes, and Alzheimer are extracted as typical caution-related diseases. The disease predicting unit 208 may identify a disease having a probability of developing in the top three after a predetermined year, for example, 20 years, as a cautionary disease, or develop when a predetermined age, for example, 65 years old is reached. Diseases that are within the top three are most likely to require attention.

  Based on the disease risk information 220 shown in FIG. 9, the disease prediction unit 208 displays a graph of the risk of developing a disease requiring attention according to age. By referring to the disease prediction screen 230 shown in FIG. 12, the user P gradually increases the risk of becoming diabetic if he / she continues the current diet, and the risk of becoming a stroke increases after the age of 50, Furthermore, we recognize that the risk of suffering from Alzheimer's increases rapidly after age 60.

  There are plenty of advice on food and life in the world, but people will not change their habits unless they recognize it as their own risk. Since the disease prediction device 104 quantitatively and rationally indicates the cautionary illness and the magnitude of the risk based on the user's eating habits, it is easy to raise awareness of the user's health risk. Such disease prediction makes it easier to increase the user's health consciousness (risk recognition) and promote the user's spontaneous eating habits.

  When the user touches the instruction button 232, a meal instruction screen is displayed.

FIG. 13 is a screen diagram of the meal guidance screen 240.
The meal instruction screen 240 is generated by the meal instruction unit 210 and transmitted to the client terminal 106. Since the user P has a risk of stroke or the like, the meal guidance unit 210 performs meal guidance for suppressing the stroke or the like. As for the meal guidance information, a fixed sentence is prepared in advance for each disease. In addition, the content of the meal guidance information is determined while taking into account the feeding information 140 of the user P. For example, since the user P has a large amount of salt intake, the standard sentence corresponding to such a user “Recommended to refrain from salt first” is included in a part of the meal guidance information. The disease prediction unit 208 may present foods and nutrients that are likely to cause etiology with a high degree of food contribution and nutrition contribution, and may call the user's attention.

  According to the feeding information 140 of the user P, it is assumed that the user P likes pickles such as “Takuan”. And if you know that salt can have an adverse effect on your stroke, you may be advised that “it seems like you often eat pickles, but you need to devise things like shallow pickles or reduce the amount and frequency of eating”. It becomes possible. The meal guidance unit 210 can suggest what to do and what to refrain from based on the eating habits of the user P (feeding information 140).

  The meal guidance unit 210 further proposes to take dietary fiber (disease-suppressing nutrients) effective for stroke prevention. The meal guidance unit 210 proposes the recommended food extracted in S30 of FIG. When the user P is allergic, the food subject to allergy is excluded from the recommended food.

  The meal guidance unit 210 may include not only the recommended food but also a menu using the recommended food in the meal guidance information. For example, when ginseng and kidney beans are disease-suppressing foods for stroke, a recipe of “carrot and kidney beans with sesame seeds” using both carrots and kidney beans may be displayed together with the cooking image. In the data storage unit 198, menu information that associates a combination of food and menu is stored in advance, and the meal guidance unit 210 includes a recommended food (disease-suppressing food) and a menu that contains as little disease-promoting food as possible. Propose.

  The life guidance unit 212 may further make a suggestion for reducing the risk of illness in terms of life other than meals. For example, various life guidance information such as exercise proposals such as walking, swimming, jogging, and stress relieving methods may be proposed to suppress stroke. In the case of the user P, although the risk of stroke is high in the future, the risk of diabetes is higher for the time being (see FIG. 12). In this case, the advice for suppressing diabetes risk may be given priority over the advice for measures against stroke, and the dietary guidance and lifestyle guidance may be given.

  Moreover, you may change the style of meal guidance and life guidance according to a risk. For example, if the risk within 5 years is less than 3%, give a soft instruction such as “Recommend” or “What should I do”, and if it exceeds 3%, “I should” or “Do” Instructional instruction may be given. The meal guidance unit 210 and the life guidance unit 212 may change not only the style but also the typeface, font size, and text (standard text) according to the disease risk.

  In life guidance, jogging is recommended for young and healthy people, and walking or pool walking may be recommended for elderly people. As described above, the contents of the dietary guidance and the life guidance may be changed based on not only the disease requiring attention and the feeding information 140 but also the user attribute and the magnitude of the morbidity risk.

FIG. 14 is a screen diagram of the inspection proposal screen 250.
The inspection proposal screen 250 is generated by the inspection proposal unit 214 and transmitted to the client terminal 106. The examination suggestion screen 250 is switched and displayed after the meal instruction screen 240. The three important illnesses of user P are stroke, diabetes, and Alzheimer. The examination proposing unit 214 selects an examination that is effective in early detection of these signs of caution and proposes it on the examination proposal screen 250.

  In general, an optional examination can be attached to a medical checkup, subject to payment of an additional fee. But if you don't know what health risks you have, you can't choose the right test. In the present embodiment, the disease predicting unit 208 identifies a cautionary disease, and the test proposing unit 214 proposes a test associated with the identified cautionary disease in advance.

  Since the user P has a risk of stroke, a head MRI (Magnetic Resonance Imaging) and a head CT (Computed Tomography) that are associated with the stroke in advance have been proposed. A blood test associated with diabetes has also been proposed. The inspection proposal unit 214 also proposes a standard fee for each inspection. The user P can select an appropriate inspection from a plurality of proposed inspections according to the budget.

In the above, the disease prediction system 100 was demonstrated based on embodiment.
The etiology analyzer 102 can search for disease-suppressing foods and disease-promoting foods based on the facts and figures of so-called big data obtained from a large amount of feeding information 140 and morbidity information 150. For this reason, unexpected knowledge about the disease may be obtained. For example, if the knowledge that “a person who likes curry is not likely to suffer from Alzheimer” is obtained, studying ingredients contained in curry powder may lead to the development of Alzheimer's preventives.

  If disease-promoting foods with particularly high food contributions to diseases or disease-suppressing foods with particularly low food contributions can be identified, it becomes easier to narrow down the selection items in the questionnaire. When there are a large number of food items to be selected, there is a possibility that the work of providing feeding information in response to a questionnaire becomes complicated. The questionnaire setting unit 213 can rationally reduce questionnaire options by removing foods that are less likely to affect diseases from selection items based on the degree of food contribution. For example, for a user who wants to check only the risk of stroke, the questionnaire setting unit 213 may exclude food items that do not significantly affect the stroke from the selection items.

  According to the disease prediction apparatus 104 of this embodiment, the risk of a plurality of types of diseases can be checked with one questionnaire. Further, the disease prediction service by the disease prediction device 104 can be received anonymously. The user can easily know the disease risk by providing the feeding information 140 and the user ID to the disease prediction device 104. With the disease prediction service as a special feature, it becomes easier to collect enormous feeding information on who is eating what. For example, by collecting the feeding information 140 of Japanese people in their twenties, it is possible to predict what disease patients will increase in the future. In this way, it is possible to predict not only an individual disease risk but also a statistical disease risk. In addition, it is useful for marketing to obtain data on what kind of food Japanese people in their 20s prefer.

  The user can always know the future disease risk by providing the feeding information 140 periodically. Further, if the risk of illness is recognized and the eating habits are improved in accordance with the meal guidance, the risk shown on the illness prediction screen 230 gradually decreases. By improving the eating habits, it is possible to visually recognize that the risk of future illness is suppressed, so that it is easy to evoke the user's desire to improve eating habits. For example, if a user who likes “cup ramen” quits “cup ramen”, the risk of “stroke” on the disease prediction screen 230 gradually decreases. If you can enjoy the process of reducing the risk of stroke, the pain (craving sensation) caused by refraining from “cup ramen” will be considerably suppressed.

  Menu proposal based on meal guidance information is also possible. In a restaurant, it is also possible to obtain recommended foods for each user from the disease prediction apparatus 104 and propose a menu including recommended foods. In general, it is said that eating out alone is bad for your health. According to the disease predicting apparatus 104, a custom-made menu that takes into account the health of the user can be proposed easily, and therefore, a business that provides “healthy eating out” is also possible.

  Furthermore, the disease prediction apparatus 104 can propose an optional test according to the health risk of the user. Unnecessarily increasing the number of optional tests also contributes to an increase in medical costs. On the other hand, not receiving a test is not preferable because it increases the risk of disease. Since the disease prediction device 104 can narrow down and propose examinations suited to the user, the disease prediction device 104 can reasonably meet the two purposes of medical cost reduction and early detection of disease.

  In addition, this invention is not limited to the said embodiment and modification, A component can be deform | transformed and embodied in the range which does not deviate from a summary. Various inventions may be formed by appropriately combining a plurality of constituent elements disclosed in the above embodiments and modifications. Moreover, you may delete some components from all the components shown by the said embodiment and modification.

  Although it has been described that the disease predicting system 100 is configured by the etiology analyzer 102 and the disease predicting device 104, the etiology analyzer 102 and the disease predicting device 104 may be configured as one unit, or one of the functions of the etiological analyzer 102. The unit may be realized by the disease prediction device 104, or a part of the function of the disease prediction device 104 may be assigned to the etiology analyzer 102. Further, a third device other than the etiology analyzer 102 and the disease prediction device 104 may take part of the function of the disease prediction system 100.

  In the present embodiment, the etiology model 160 has been described as a neural network model. The pathogenesis model 160 does not have to be one type, and a plurality of types may be prepared. For example, multiple types of etiology models 160 with different numbers of intermediate layers may be prepared. In addition to logistic sigmoid functions, multiple types of activities such as hyperbolic tangent functions, normalized linear functions, etc. A plurality of types of etiology models 160 corresponding to the conversion function may be prepared. Of the plurality of pathogenesis models 160, the pathogenesis model 160 having the highest prediction accuracy may be employed. As is known, there is also a method for improving the prediction accuracy by averaging the outputs of a plurality of pathogenesis models 160.

  The setting of the etiology model 160 may be changed for each disease or for each user attribute. For example, when predicting a stroke in a man in their 30s, the etiology model 160 having n intermediate layers and an activation function of a logistic sigmoid function is selected. When predicting a heart disease in a woman in their 40s, m intermediate layers are selected. The pathogenesis model 160 using the tangent function as an activation function may be selected.

  In the present embodiment, the etiology model 160 is described as being formed by a neural network. However, the etiology model 160 may be formed by a multivariate analysis model other than the neural network such as multiple regression analysis or discriminant analysis.

  In the present embodiment, it has been described that the etiology model 160 is prepared for each disease and age group. In addition to this, a plurality of pathogenesis models 160 may be prepared based on other user attributes such as sex and residential area.

  In the present embodiment, the disease prediction device 104 identifies a disease requiring attention based on the food contribution degree to the disease and the feeding information 140. As a modification, etiology information 170 defining the degree of nutritional contribution to the disease may be prepared. The disease prediction unit 208 may identify a nutrient contained in the food ingested by the user, and may identify a disease having a positive correlation with the nutrient as a disease requiring attention.

  The etiology model 160 has been described as a model that associates the relationship between food and disease. As a modification, the etiology model 160 may be formed as a model for associating the relationship between nutrients and diseases.

  Nutrients included in the food or meal menu and their contents may be registered in advance in a database, and nutrient data may be provided to the user.

  Dedicated software for inputting the feeding information 140 to the etiology analyzer 102 or the disease predictor 104 may be installed in the client terminal 106. For example, the client terminal 106 may take an image of a meal, specify a food from the meal image, and automatically transmit the food information 140. In the client terminal 106, not only the food contained in the meal but also the nutrient contained in the food may be specified.

  The hospital 108 may transmit the morbidity information 150 of the user Q to the etiology analyzer 102 when an abnormal value is detected in a health examination of a certain user Q. At this time, the etiology analyzer 102 may correct the etiology model 160 based on the feeding information 140 and the disease information 150 of the user Q.

  Nutrients may be defined as more comprehensive nutrients such as “carbohydrates, lipids, carbohydrates, minerals, vitamins, dietary fiber” as well as vitamin A, folic acid, and salt. Further, the nutrient is not necessarily “nutrition” such as a food additive, and can be a target as long as it is at least a component contained in the food.

  In the present embodiment, it has been described that the pathogenesis is investigated by connecting the correlation between the feeding information 140 and the morbidity information 150 by the pathogenesis model 160 and the disease is predicted. In addition to meals, mental trends determined by psychological tests (eg, positive and negative personality), physical information such as height and weight, gender, residential area, race, age, history of the person or parent, etc. May be included in the input data of the etiology model 160. By performing such an analysis, for example, it is possible to analyze the etiology other than meals, such as women are less likely to get lung cancer, and residents in warm regions are less likely to get strokes.

  The “disease” to be predicted in the present embodiment is not limited to serious illnesses such as stroke and heart disease, but may be defined as a concept including health abnormalities such as depression, headache, hay fever and poor physical condition.

  DESCRIPTION OF SYMBOLS 100 Disease prediction system, 102 Etiology analysis apparatus, 104 Disease prediction apparatus, 106 Client terminal, 108 Hospital, 110 Internet, 120 User interface processing part, 122 Data processing part, 124 Communication part, 126 Data storage part, 128 Input part, 130 Output unit, 132 etiology information output unit, 134 contribution calculation unit, 136 feeding information acquisition unit, 138 disease information acquisition unit, 140 feeding information, 150 disease information, 160 etiology model, 170 etiology information, 180 nutrition information, 190 User interface processing unit, 192 communication unit, 196 data processing unit, 198 data storage unit, 200 input unit, 202 output unit, 204 feeding information acquisition unit, 206 question transmission unit, 208 disease prediction unit, 210 meal guidance unit, 212 Life Guidance Department, 213 Questionnaire setting unit, 214 test proposed unit, 220 disease risk information, 230 disease expected screen 232 teaching button, 240 dietary advice screen 250 test proposal screen

Claims (18)

A feeding information acquisition unit for acquiring feeding information indicating the food ingested by the user;
A disease information acquisition unit for acquiring disease information indicating a disease affected by the user;
A contribution calculation unit that calculates the magnitude of the correlation between the disease and the food as the food contribution by referring to the feeding information and morbidity information of each of the plurality of users,
An etiology analyzer comprising: a disease and an etiology information output unit that outputs the degree of food contribution of each of a plurality of types of food to the disease as etiology information.   The contribution calculation unit inputs a plurality of types of foods ingested by the user, and uses a multivariate analysis model that outputs the diseases affected by the user after eating, and the food contributions of each of the plurality of types of foods to the diseases The etiology analyzer according to claim 1, wherein the etiology is calculated backward. The contribution calculation unit further refers to nutrition information indicating the nutrients contained in the food and the content thereof, and calculates the magnitude of the correlation between the disease and the nutrient as the nutrition contribution,
The etiology analyzer according to claim 1 or 2, wherein the etiology information output unit further outputs a disease and a nutrient contribution degree of each of a plurality of types of nutrients for the disease as the etiology information. A feeding information acquisition unit for acquiring feeding information indicating the food ingested by the user;
A pathology information that defines the degree of food contribution indicating the degree of correlation between the disease and food, and a disease prediction unit that refers to the feeding information and predicts a disease that is likely to cause the user. A disease prediction device. A question transmission unit that transmits questionnaire information for selecting and inputting the food taken by the user;
The disease prediction apparatus according to claim 4, wherein the feeding information acquisition unit acquires feeding information based on the user's answer to the questionnaire information.   The disease prediction apparatus according to claim 4, wherein the feeding information acquisition unit acquires user feeding information from meal image information.   The feeding information acquisition unit acquires the feeding information of the user based on the menu of the user's meal by referring to the menu information that associates the menu and the food included in the menu. The disease prediction apparatus according to 5 or 6.   The meal guidance part which refers to the said etiology information and presents the food which has a negative food contribution with respect to the said estimated disease as a recommended food, In any one of Claim 4 to 7 characterized by the above-mentioned. The disease prediction apparatus as described. In the etiology information, the magnitude of the correlation between the disease and the nutrient is further defined as the degree of nutrition contribution,
The dietary guidance unit refers to the etiology information and presents a food containing a nutrient having a negative nutritional contribution to the predicted disease as a recommended food. Disease prediction device.   The meal guidance unit refers to the user's allergy information, and excludes a food subject to allergy from a recommendation target among one or more foods having the predicted disease and negative food contribution. The disease prediction apparatus according to claim 8 or 9.   The meal guidance part which presents the menu containing the foodstuff which has the negative food contribution with respect to the said disease with reference to the said etiology information, The food guidance part characterized by the above-mentioned. The disease prediction apparatus as described.   The disease prediction apparatus according to any one of claims 4 to 11, further comprising a life guidance unit that outputs life guidance information that is a proposal for life improvement in response to the predicted disease. .   The disease prediction according to claim 12, wherein the life guidance unit changes the life guidance information according to a difference between an age at which the predicted occurrence of the disease is predicted and a current age of the user. apparatus.   The disease prediction apparatus according to any one of claims 4 to 13, further comprising a test proposal unit that proposes a menu of health tests in response to the predicted disease. A feeding information acquisition unit for acquiring feeding information indicating the food ingested by the user;
Disease prediction that predicts a disease likely to be affected by the user based on the nutrients contained in the food consumed by the user with reference to the etiology information that defines the degree of nutritional contribution indicating the magnitude of the correlation between the disease and the nutrient A disease prediction apparatus comprising: a unit. Equipped with etiology analyzer and disease predictor,
The etiology analyzer is
A feeding information acquisition unit that acquires feeding information indicating the food ingested by the first user;
A disease information acquisition unit for acquiring disease information indicating a disease affected by the first user;
A contribution calculation unit that calculates the magnitude of the correlation between the disease and the food as the food contribution by referring to the feeding information and morbidity information of each of the plurality of users,
An etiology information output unit that outputs the degree of food contribution of each of the plurality of types of food to the disease as etiology information,
The disease prediction device
A question transmission unit that transmits questionnaire information for selecting and inputting the food ingested by the second user;
A feeding information acquisition unit that acquires the feeding information of the second user based on the answer of the second user to the questionnaire information;
A disease prediction unit that refers to the etiology information and the second user's feeding information and predicts a disease that is likely to cause the user;
A disease prediction system, comprising: a questionnaire setting unit that updates questions of the questionnaire information according to the food contribution degree. A function for obtaining eating information indicating foods consumed by the user;
A function of acquiring disease information indicating a disease affected by the user;
A function for calculating the degree of correlation between a disease and food as a food contribution by referring to feeding information and morbidity information of each of a plurality of users,
A pathogenesis analysis program characterized by causing a computer to exhibit a disease and a function of outputting food contribution degrees of each of a plurality of types of foods to the disease as pathogenesis information. A function for obtaining eating information indicating foods consumed by the user;
Let the computer demonstrate the pathogenesis information that defines the degree of food contribution indicating the magnitude of the correlation between the disease and food, and the function of predicting the disease that the user is likely to be affected by referring to the user's feeding information A disease prediction program characterized by this.

Images