A health information intelligent recommendation system and management method based on big data

Abstract

The present invention discloses a health information intelligent recommendation system and recommendation method based on big data. The recommendation system includes a client and a management server, wherein the client receives a user's health information retrieval request and sends the request to the management server through a network. The management server is provided with a health database and a health information search engine, receives the health information retrieval request sent by the client and controls the health information search engine to perform matching retrieval in the health database, and returns the matching health data to the client. The present invention constructs a health information search engine and a TCM constitution test database, and a dietary health database based on TCM theory (a database of ingredients and a recipe database with TCM attributes), and accurately recommends a diet plan for conditioning the body according to the user's TCM constitution, chronic diseases / common diseases, and special populations, intelligently displays the judgment of the food ingredients and recipes, and realizes intelligent and humanized information recommendation. Users can achieve this by installing an APP on a mobile terminal or through applications such as WeChat applets, which is very convenient to use.

Description

Technical Field

[0001] The present invention belongs to the field of health management technology, and in particular relates to a management system and method for providing healthy diet and personalized customized meals based on traditional Chinese medicine big data. Background Art

[0002] At present, in terms of dietary health management, the health management systems on the market are all based on disease risk management, follow-up management, chronic disease management and specific population management based on Western medicine theory. As for diet, they only provide some simple information, such as menus and food introductions. This information is not actually combined with the user's health status and only stays at the level of information supply and demand, such as:

[0003] 1) Failure to connect with the user's physical condition, especially the traditional Chinese medicine constitution, such as whether these foods and recipes are suitable for people with these constitutions;

[0004] 2) Failure to connect with the user's physical disease information, such as whether these foods and recipes are suitable for people with these diseases;

[0005] 3) Failure to connect with users in special groups, that is, when users are in special groups, what foods are suitable for them to eat and good for their health;

[0006] 4) Failure to connect with the current solar terms, that is, what foods are suitable to eat during the current solar terms and are beneficial to health;

[0007] 5) Failure to bring out the TCM functional properties of food ingredients, that is, how can sub-healthy groups and sick people choose food ingredients that are suitable for them, how can they eat to avoid getting sick, and how can they eat to stay healthy;

[0008] 6) Failure to recommend healthy foods suitable for the individual based on his or her health condition;

[0009] 7) The lack of a strong food knowledge base system, especially the knowledge base of the TCM properties of food ingredients, also prevents the general public from fully enjoying the benefits brought by the TCM theory system that has been passed down for thousands of years;

[0010] In summary, the prior art has not yet developed a management system and method for recommending a personalized diet based on the user's personal TCM constitution type, disease information, special population group and other information for dietary intervention and health care. Summary of the invention

[0011] In view of the defects in the prior art, the purpose of the present invention is to provide a health information intelligent recommendation system and method based on big data, which can intelligently and comprehensively solve the user's dietary and health problems. Users can customize their diet according to their own TCM constitution characteristics and their own health conditions to achieve a peaceful and healthy state and the goal of health management; at the same time, it promotes the development of the TCM system towards digitalization and intelligence.

[0012] In order to achieve the above purpose, the technical solution adopted by the present invention is:

[0013] A health information intelligent recommendation system based on big data, comprising:

[0014] The client is used to receive a user's health information retrieval request and send the request to the management server through the network;

[0015] The management server is provided with a TCM constitution test database, a health database (a TCM food material database and a recipe database), a dietary mall platform and a health information search engine. When receiving a health information search request sent by the client, the management server is used to control the health information search engine of the management server to perform a matching search in the health database and return the matching health data to the client through the network;

[0016] In the process of processing the client's retrieval request, the management server first receives an initial retrieval request from the client, retrieves retrieval results matching the initial retrieval request from the dietary health database, the disease database, and the traditional Chinese medicine constitution test database, and returns the retrieval results to the client user; when a secondary retrieval request indicating a selected retrieval result from the retrieval results is received from the client, the management server generates a matching link, and the matching link is used to identify the hit database from which the selected retrieval result is retrieved; the management server uses the matching link to link the client to the hit database identified in the matching link, so as to process the secondary retrieval request in the hit database.

[0017] According to the preferred technical solution, the dietary health database is based on the TCM constitution theory and the basic theory of TCM. According to the causes and symptoms of biased constitution, it collects all common food ingredients, Chinese medicinal materials and food and medicine substances with the same origin as medicine, including food ingredient management and recipe management, wherein food ingredient management includes food ingredient type, TCM properties of food ingredients (nature, flavor and meridians), main efficacy of food ingredients, nutritional components of food ingredients, and food taboos; the recipe management includes the TCM efficacy of recipes, TCM properties of recipes, recipe cooking methods, and recipe taboos; wherein the taboos include analysis of the reasons for the taboos of food ingredients (or recipes) for TCM constitution, common diseases, chronic diseases, and special populations, and also includes the nutritional content of food ingredients and recipes suitable for different solar terms.

[0018] According to the preferred technical solution, the dietary mall platform includes a personalized dietary mall platform, and the dietary mall platform includes a food library, food dimensions, food details, and food suitable for different physical conditions.

[0019] According to the preferred technical solution, the client is provided with a health information receiving unit, a disease information receiving unit and a special population setting unit; the health information receiving unit receives data input by the user, makes a judgment based on the data, and outputs the user's TCM constitution type; the disease information receiving unit allows the user to select the disease type to which he belongs, and the special population setting unit allows the user to select the special population to which he belongs, and the management server outputs health data suitable for the user to the client based on the user's TCM constitution type, disease type, and special population to which he belongs.

[0020] According to the preferred technical solution, the management server receives the user's health risk data and user history data, processes the health risk data by using an unsupervised learning algorithm to generate a fixed-size vector representation, extracts health risk features from the vector representation by using a multi-layer convolutional neural network, determines the dietary conditions associated with the health risk features, generates a dietary recommendation plan based on the dietary conditions and user history data, and displays the dietary recommendation plan to the user.

[0021] According to the preferred technical solution, after returning the matched health data to the client, the user's interest feedback information on the health data is introduced into the matching filtering algorithm, a similar user set is returned to the user, and recommendations are made to the current user based on the interest feedback information of the similar user set;

[0022] Assume that the number of interest feedback information of health recommendation results for users u and v is N, and the interest feedback information of user u is R u =(r u1 ,r u2 ,…,r uN ), user v’s interest feedback information Rv=(r v1 ,r v2 ,…,r vN ), r ui and r vi are the feedback information of user u and v on the i-th health recommendation result, i.e., the matching search item. The feedback information can be expressed as satisfaction or interest. The difference of feedback information of user u and v can be expressed as:

[0023] DELTA(u,v)=|R u -R v |

[0024] The common interest feedback information difference DELTA (u, v) is dimensionally counted. Assuming there are n dimensions, its frequency distribution is expressed as:

[0025] F(DELTA(u,v))=(p 1 ,p 2 ,…,p n )

[0026] Among them, p 1 ,p 2 ,…,p n represents the frequency of occurrence of each dimension, so the difference of F(DELTA(u,v)) is expressed as:

[0027] H(F(DELTA(u,v)))=-

[0028] In order to consider the difference of interest feedback information, the distribution probability is p i Interest feedback information difference |DELTA(p i )|Add to the above formula to get:

[0029] sim H (u,v)=-×|DELTA(p i )|

[0030] Calculate the weighted difference similarity sim WH (u,v):

[0031] sim WH (u,v)=sim H (u,v)·f(u,v)

[0032] f(u,v) is a weighted function, N is the number of common feedback information of users u and v, and f(u,v) is taken as:

[0033] f(u,v)=1-

[0034] Among them, the value range of α is 0<α<1;

[0035] Personalized recommendations are made based on the interests of each user. Let the interest vector of user u be

[0036] Ist u =(I u,1 ,I u,2 ,…,I u,n )

[0037] I u,i =c u,i / c u

[0038] Among them, I u,irepresents the interest vector of user u for the i-th recommendation result, c u,i is the number of recommended results for the i-th item whose interest level of user u is higher than the mean interest level, c u Represents the total number of all recommended results with a higher interest level than interest. The similarity calculation process of the interests of two users is as follows:

[0039] simIst(u,v)=Jac(Ist u ,Ist v )

[0040] Jac is the Jaccard similarity coefficient operation; through the similarity Sim(u,v) between users, the closest neighbor set v is determined for user u;

[0041] Sim(u,v)=λ·sim WH (u,v)+(1-λ)·simIst(u,v)

[0042] Where λ is the adjustment parameter;

[0043] According to the interest feedback information of the similar user set, an optimized recommendation is made to the current user. In another aspect, the present invention also provides a health information intelligent recommendation method based on the aforementioned health information intelligent recommendation system, comprising the following steps:

[0044] When the client receives a user's health information retrieval request, the client sends the request to the management server;

[0045] The management server receives the user's health information search request and calls the health information search engine according to the user's request;

[0046] The health information search engine performs engine data retrieval in the dietary health database; the health information search engine returns the retrieved matching data to the client.

[0047] According to the preferred technical solution, the big data-based health information intelligent recommendation method further includes:

[0048] The management server receives the user's health information retrieval request, and first normalizes the request according to a predefined data format, specifically dividing the request data format into two parts: a param part and a data part; the param part includes an MD5 signature for security verification; the data part is the input parameter data required for functional logic processing; the normalized data is input into the service platform entrance.

[0049] According to the preferred technical solution, the big data-based health information intelligent recommendation method further includes:

[0050] After the data is input into the service platform entrance, the parameters are converted and checked. The parameters include the service number serviceid, the data format dataformat, the user unique identifier sessionId, and the signature string MD5. The parameter conversion is performed through a converter, and the parameters are converted from the " / / ; / / " and " / / = / / " formats into the framework flow object Param object; during the conversion process, the data is verified for validity, including format verification, service number verification, and MD5 signature verification, and then service routing is performed according to the service number serviceid.

[0051] According to the preferred technical solution, the big data-based health information intelligent recommendation method further includes:

[0052] The data after service routing is input into the pre-filter pool, which includes a decompression filter, a decryption filter, a remote service check filter, and a login filter. The filter pool also supports subsequent dynamic expansion. Only when all the checks of the pre-filter pool are passed, the retrieval service module of the health information search engine is called for logical processing.

[0053] According to the preferred technical solution, the health information search engine in the big data-based intelligent recommendation method for health information is used to perform keyword segmentation processing, and engine data retrieval is performed in the health database based on the segmentation results. The health information search engine performs traversal queries, and during the query process, a matching score is given to the information input by the user, thereby feeding back the valid search information set to the user end, and arranging the information in descending order according to the matching score; these matching data are then encapsulated, that is, the part of the data is unified into the currently common JSON format.

[0054] According to the preferred technical solution, the big data-based health information intelligent recommendation method further includes:

[0055] The encapsulated JSON format data is input into a post-filter, which includes an encryption filter, a compression filter, and a result format encapsulation filter. The data output by the post-filter is routed via the service and then returned to the client via the service platform entrance.

[0056] The present invention is particularly intelligent and user-friendly by constructing a powerful health information search engine and a TCM constitution test database and a dietary health database (a TCM food ingredient database and recipe database); users only need to enter the name of the ingredient or the recipe name, and the system will intelligently display the pros and cons of the ingredients and recipes based on the user's TCM constitution, chronic diseases / common diseases, and special populations, and display the pros and cons of the ingredients or recipes that automatically match the user's TCM constitution, special populations, chronic diseases, and common diseases, and analyze the efficacy and nutritional composition of the food. Users can achieve this by installing the APP on the mobile terminal, which is very convenient to use. The specific beneficial effects are as follows:

[0057] (1) The present invention constructs a dietary health and dietary mall platform database based on TCM theory, which will store all common food information, recipe information, disease information, TCM constitution information, special population information, dietary commodity information, which is comprehensive and has strong support;

[0058] (2) The present invention combines the dietary health database with the TCM attribute function of food ingredients to solve the problem of what people in sub-health state should eat to avoid getting sick and how to eat to stay healthy; the dietary health database is connected with people's TCM constitution, that is, to solve the problem of whether a certain type of constitution is suitable or not suitable for certain food ingredients and what kind of recipes are suitable; the dietary health database is connected with people's physical diseases, that is, to solve the problem of whether a certain type of physical disease is suitable or not suitable for certain food ingredients and what kind of recipes are suitable; the dietary health database is connected with special groups of people, that is, to solve the problem of whether certain food ingredients are suitable or not suitable for people at different stages and what kind of recipes are suitable; the dietary health database is connected with the current solar term to find out what food is suitable for the current solar term and what is good for the body;

[0059] (3) The present invention connects the dietary mall platform with the individual's TCM constitution, identifies the body and provides diet, solves the problem of personalized food customization, and independently recommends the best individual food; it integrates health preservation and disease treatment into daily life and diet, and gives full play to the positive role of food.

[0060] (4) After receiving the user's request, the management server of the present invention first normalizes the request according to a predefined data format, thereby solving security issues, simplifying the processing logic of the front-end input parameters, and unifying the verification method of the validity of the back-end input parameters; thus making subsequent data processing faster.

[0061] (5) The present invention converts and checks the necessary parameters at the entrance of the service platform, and the role of the pre-filter pool makes the retrieval data safe and accurate, reduces the pressure on the health information search engine, and makes the retrieval speed of the health information search engine faster and more accurate. In conjunction with the role of the post-filter pool, the result data format is unified, which is convenient for system integration and processing.

[0062] (6) After the results of the present invention are transformed into production, they can effectively divert and digest Chinese medicinal materials, food ingredients, and agricultural and sideline products, which is conducive to promoting the development of agricultural industrialization, improving the economic benefits of growers, and empowering rural revitalization. Through the application of this system, it is also possible to trace and manage agricultural products, promote the modernization and internationalization of the traditional Chinese medicine industry, and contribute to the inheritance and development of traditional Chinese medicine culture. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0064] Figure 1 is a schematic block diagram of a health information intelligent recommendation system based on big data according to the present invention;

[0065] Figure 2 It is a schematic flow chart of the big data-based intelligent recommendation method for health information according to the present invention. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0067] The embodiments of the present invention are further described in detail below with reference to the accompanying drawings.

[0068] The present invention provides a health information intelligent recommendation system based on big data, such as Figure 1 As shown, it includes a client and a management server, the management server is provided with a health database, a TCM constitution test database, and a health information search engine, the client is used to receive a user's health information search request and send the request to the management server, the management server is used to receive the health information search request sent by the client and control the health information search engine to perform a corresponding search in the health database and return the corresponding data to the client. The client can be an Android client or an iOS client.

[0069] The client is provided with a health information receiving unit for receiving information related to the user's TCM constitution. The health information receiving unit can be a set of TCM constitution test interactive interfaces, which include a series of questions and answers. The user makes a choice based on his or her own situation. The system will derive a comprehensive score based on the answer selected by the user, and then divide the constitution based on the score. The client is provided with a disease information receiving unit for setting the user's disease. The user can register a login account and selectively set his or her own disease information according to his or her health status. The client is provided with a special population setting unit for setting special populations of the user. The user can register a login account and set the special population to which he or she belongs.

[0070] The health database includes a dietary health database and a dietary mall platform. The dietary health database includes food material management and recipe management, wherein food material management includes food material type, food material TCM properties (nature and flavor meridians), food material main efficacy, food material nutritional ingredients, and food material taboos. The food material types include: fruits, vegetables, beans, edible fungi, algae, nuts, and seeds, etc.; the food material efficacy includes: replenishing qi and strengthening spleen, moistening intestines and relieving constipation, clearing away heat and detoxifying, promoting blood circulation and menstruation, diuresis and swelling, etc.; the food material nutritional ingredients include: such as carbohydrates, fats, proteins, vitamins, trace elements, etc.; the food material taboos include a specific analysis of the appropriate reasons and taboo reasons for food materials for TCM constitutions, diseases, and special populations. For example: fennel has the effects of strengthening the stomach, dispersing cold, promoting qi, relieving pain, and assisting yang. It mainly treats abdominal pain and other symptoms caused by cold in the middle jiao. Fennel is suitable for consumption for yang deficiency constitution, but not for yin deficiency constitution. Recipe management includes recipe efficacy, recipe attributes, recipe cooking methods, and recipe taboos. The recipe effects include: such as stimulating appetite and digestion, losing weight and removing fat, strengthening the spleen and removing dampness, nourishing blood and removing wind, etc.; the recipe attributes include: such as meat dishes, vegetarian dishes; main ingredients, auxiliary ingredients, and side dishes, etc.; the recipe cooking methods include: such as stir-frying, frying, sticking, cooking, deep-frying, braising, boiling, stewing, simmering, baking, steaming, braising, sauce, smoking, roasting, boiling, pickling, mixing, pulling silk, and fanning, etc.; the recipe taboos include a specific analysis of the reasons why the recipes are suitable and the reasons why they are taboo for traditional Chinese medicine constitutions, diseases, and special groups of people.

[0071] The dietary mall platform includes a personalized dietary mall platform, and the dietary mall platform includes a food library, food dimensions, food details, and food suitable for traditional Chinese medicine constitution. The food dimensions include fast food, dry goods, beverages, grains and oils, fruits and vegetables, etc.

[0072] In the process of processing the retrieval request of the client, the management server first receives an initial retrieval request from the client. The retrieval results matching the initial retrieval request are retrieved from the dietary health database, the disease database, and the traditional Chinese medicine constitution test database, and the management server returns the retrieval results to the client user. When a secondary retrieval request indicating a selected retrieval result in the retrieval results is received from the client, a matching link is generated by the management server, and the matching link is used to identify the hit database from which the selected retrieval result is retrieved. The management server uses the matching link to link the client to the hit database identified in the matching link to process the secondary retrieval request in the hit database.

[0073] Optionally, multiple search parameters included in the user query are identified, and multiple query statements representing the multiple search parameters in the user query are determined by parsing the user query to generate a parse tree with multiple nodes, wherein each node represents a query statement among the multiple query statements; since the types of queries may include single-database queries and cross-database queries, single-database query statements of multiple query statements are identified, wherein the single-database query statements are query operations executed separately by a single hit database; combined query statements of the multiple query statements are identified, wherein the combined query operation is a cross-database query statement that cannot be executed by a single database; after the single-database queries are executed separately by multiple hit databases, single-database query results are received; and the cross-database query is executed based on a set of single-database query results.

[0074] The big data-based intelligent recommendation system for health information of the present invention includes three platforms, namely: a basic platform, a business platform and a mobile application platform, wherein the entire basic platform runs through the SOA modular development concept and the ESB server bus concept, and as the basis of the business platform and the mobile application end, the basic platform provides services including message push, SMS service, session management, cache management, log management, service management, service routing, service registration, startup, uninstallation, etc. The business platform opens interface calls to the mobile end, and the interfaces are divided into security control interfaces and non-security control interfaces. The security control interface first needs to check and verify user permissions to ensure that this type of interface is not invaded by other illegal applications or individuals, thereby ensuring data security. The mobile application platform includes a health assessment platform, a dietary guidance platform, an inspection report platform, an exchange and sharing platform, and a dietary mall platform.

[0075] In order to identify health risks and determine health information recommendations, in another embodiment of the present invention, the management server provides accurate dietary plan recommendations by matching health risk data, dietary taboos and user preferences.

[0076] First, the management server receives the user's health risk data and user history data, processes the health risk data by using an unsupervised learning algorithm to generate a fixed-size vector representation, extracts health risk features from the vector representation by using a multi-layer convolutional neural network, determines the dietary conditions associated with the health risk features, generates a dietary recommendation based on the dietary conditions and the user's history data, and displays the dietary recommendation for the user. Since health risk data, health conditions, and user preferences can have different formats and form massive data sets, in order to improve the accuracy of health risk identification, an embodiment of the present invention processes user data through a deep learning engine that can use similarities between users to extract individual information.

[0077] The deep learning engine input data includes trained health risk data, dietary conditions, and health risk data to be trained. The trained health risk data and the health risk data to be trained are vectorized in the trained feature vector and the feature vector to be trained, respectively. The vectorized trained health risk data and various dietary conditions are processed by the deep learning engine to identify health recommendation schemes based on dietary taboos and health risk matching. Specifically, the health risk data is processed to generate a phrase containing multiple terms, the phrase is formatted to generate a matrix, constraints are set for each term of the matrix, weights are added to each term of the matrix, and vector representations are generated based on the matrix. After the deep learning engine generates vectors from characters, the vectors are used as inputs to the clustering algorithm to extract health risk features. The prediction module receives the vectorized health risk data to be trained and the association between dietary taboos and health risks from the deep learning engine to determine the health recommendation scheme. The prediction module adjusts the recommendation results based on multiple conditions. The adjustment conditions include the similarity of various user data, etc. The prediction module then sends the health recommendation to the client.

[0078] Considering the impact of cold start in the recommendation system on the matching retrieval accuracy, the user's interest feedback information on health recommendation results is introduced into the matching filtering algorithm. Combined with the weighted similarity calculation method, a set of similar users is returned to the user, and recommendations are made to the current user based on the interest feedback information of the similar user set.

[0079] Assume that the number of interest feedback information of health recommendation results for users u and v is N, and the interest feedback information of user u is R u =(r u1 ,r u2 ,…,r uN ), user v’s interest feedback information Rv = (r v1 ,r v2 ,…,r vN ), r ui and r viare the feedback information of user u and v on the i-th health recommendation result, i.e., the matching search item. The feedback information can be expressed as satisfaction or interest. The difference of feedback information of user u and v can be expressed as:

[0080] DELTA(u,v)=|R u -R v |

[0081] The common interest feedback information difference DELTA (u, v) is dimensionally counted. Assuming there are n dimensions, its frequency distribution is expressed as:

[0082] F(DELTA(u,v))=(p 1 ,p 2 ,…,p n )

[0083] Among them, p 1 ,p 2 ,…,p n represents the frequency of occurrence of each dimension, so the difference of F(DELTA(u,v)) is expressed as:

[0084] H(F(DELTA(u,v)))=-

[0085] In order to consider the difference of interest feedback information, the distribution probability is p i Interest feedback information difference |DELTA(p i )|Add to the above formula to get:

[0086] sim H (u,v)=-×|DELTA(p i )|

[0087] Considering the number of common interest feedback information, the more the number of interest feedback information, the greater the probability that two users are similar, and the weighted difference similarity sim is calculated. WH (u,v):

[0088] sim WH (u,v)=sim H (u,v)·f(u,v)

[0089] f(u,v) is a weighted function, N is the number of common feedback information of users u and v. The larger N is, the greater the probability that two users are similar. Take f(u,v) as:

[0090] f(u,v)=1-

[0091] Among them, the value range of α is 0<α<1.

[0092] When two users have similar interest preferences, that is, they both give relatively satisfactory feedback on a certain recommendation result, it can be considered that the similarity between the two is high, so personalized recommendations are made based on the interests of each user. Suppose the interest vector of user u is

[0093] Ist u =(I u,1 ,I u,2 ,…,I u,n )

[0094] I u,i =c u,i / c u

[0095] Among them, I u,i represents the interest vector of user u for the i-th recommendation result, c u,i is the number of recommended results for the i-th item whose interest level of user u is higher than the mean interest level, c u Represents the total number of all recommended results with a higher interest level than interest. The similarity calculation process of the interests of two users is as follows:

[0096] simIst(u,v)=Jac(Ist u ,Ist v )

[0097] Jac is the Jaccard similarity coefficient operation. Through the similarity between users Sim(u,v), the closest neighbor set v is determined for user u:

[0098] Sim(u,v)=λ·sim WH (u,v)+(1-λ)·simIst(u,v)

[0099] Among them, λ is an adjustment parameter. Finally, based on the interest feedback information of similar user sets, optimized recommendations are made to the current user.

[0100] like Figure 2 As shown, the present invention also provides a method for intelligent recommendation of health information based on big data, comprising the following steps:

[0101] 1) When the client receives a user's health information retrieval request, the client sends the request to the management server; wherein the health information retrieval request may include retrieval requests for food names, recipe names, disease names, symptom names, etc.

[0102] 2) The management server receives the user's health information retrieval request and first normalizes the request according to the predefined data format, specifically dividing the request data format into two parts: the param part and the data part; the param part includes an MD5 signature for security verification; the data part is the input parameter data required for functional logic processing. By clearly dividing the input parameters into two parts, the processing logic of the front-end input parameters is simplified, and the verification method of the validity of the back-end input parameters is unified, and the normalized data is input into the service platform entrance.

[0103] 3) After the data is input into the service platform entrance, the necessary parameters (i.e., service number serviceid, data format dataformat, user unique identifier sessionId, signature string MD5) are converted and checked. The parameter conversion is performed through a converter, and the parameters are converted from the " / / ; / / " and " / / = / / " formats into a framework flow object Param object, which will be valid throughout the request cycle until the end of the request cycle; during the conversion process, the data is verified for validity, and only data that meets the requirements is considered safe, such as identifying the data format, service number, user unique identifier, etc.; the check includes format verification, service number verification, and MD5 signature verification.

[0104] 4) Perform service routing according to the service number serviceid. The routing strategies during routing include cache hit, heat cache, database loading, and first-in-first-out. The cache hit refers to searching the cache for the corresponding service number, and using it directly if it exists. The heat cache refers to caching frequently used services for reuse. The database loading includes, if the cache hit fails, loading the service from the database, and using the service if the loading is successful. The first-in-first-out includes, for the services in the cache, if the heat cache is not used, unloading the services in the cache according to the first-in-first-out rule.

[0105] 5) Input the service routed data into the pre-filter pool, which is composed of a series of filters, such as decompression filter, decryption filter, remote service check filter, login filter, etc. These filters are implemented through the aspect interception technology. The filter pool also supports subsequent dynamic expansion. Only after all the checks of the pre-filter pool are passed, the retrieval service module of the health information search engine will be actually called for logical processing;

[0106] 6) The health information search engine performs keyword segmentation, and performs engine data retrieval in the health database based on the segmentation results. The health information search engine performs traversal queries, and during the query process, it will score the matching degree with the information entered by the user, such as 100 points for a complete match, 80 points for only a partial match, and so on, so as to feed back the effective retrieval information set to the user end and sort it in descending order according to the matching degree score. These matching data are then encapsulated, that is, the system framework will unify this part of the data into the current common JSON format for subsequent docking with other partners.

[0107] 7) The encapsulated data passes through the post-filter, which includes an encryption filter, a compression filter, and a result format encapsulation filter. The data output by the post-filter is routed through the service and then returned to the client through the service platform entrance.

[0108] The client is provided with a health information receiving unit, a disease information receiving unit and a special population setting unit. The health information receiving unit receives the data input by the user, makes a judgment based on the data, and outputs the TCM constitution type of the user. The disease information receiving unit allows the user to select the disease type to which he belongs, and the special population setting unit allows the user to select the special population to which he belongs. The management server outputs the health data suitable for the user to the client based on the user's TCM constitution type, disease type, and special population to which he belongs.

[0109] In order to perform engine data retrieval in a health database based on word segmentation results, in a further embodiment of the present invention, a feature word set is extracted and then the feature words are semantically expanded to obtain target keywords, thereby determining the index of each feature word in the word segmentation result.

[0110] By calculating the frequency of all feature words in the user input text and the index in the corresponding text, we take the first k feature words with the highest frequency and record their indexes, which are recorded as M kw ={(W i , M i ), i = 1, 2, ..., k}, where W i is the i-th feature word, M i W i The index matrix of occurrences.

[0111] On this basis, through the semantic expansion process of generating keywords based on the feature word set, the target keyword containing the feature word set is obtained. Specifically, the feature word set M is first read kw , each time traversing M kw A characteristic word W in i and its corresponding index matrix M i , and marked as an intermediate node. iAs the root node, start building a splay tree. When there are unexpanded intermediate nodes in the splay tree, read each intermediate node and its index matrix separately, and then do the following:

[0112] Step 1: The length of the feature word sequence from the root node to the current intermediate node is recorded as L. If L<5, the feature words in the text corresponding to the subsequent index of each element in the index matrix of the intermediate node are placed into the intermediate set Φ i , go to step 2; if L ≥ 5, go to step 4;

[0113] Step 2: Φ i All elements are semantically expanded, where Φ i Each element in is denoted as w j , for w j The index matrix of its parent node that appears after the feature word in the text is assigned a value. If the number of elements in its index matrix is ​​less than 4, it is marked as a leaf node, otherwise it is marked as an intermediate node;

[0114] Step 3: From the intermediate set Φ i Take the next element w from j+1 , and perform step 2 again until all Φ i Elements in

[0115] Step 4: Initialize the leaf node set as the child nodes of the intermediate node, complete the processing of the intermediate node, select the next intermediate node that has not been expanded, and re-execute step 1; if all intermediate nodes are processed, execute step 5;

[0116] Step 5: Traverse a leaf node of the splay tree, and trace back from the current leaf node to the root node until the first feature word node is encountered. The feature word combination from the beginning of the root node to the end of the feature word node is used as the target keyword;

[0117] Step 6: Traverse the other leaf nodes of the splay tree in turn, iterate step 5, until all leaf nodes are processed, and finally output the target keyword set with frequency and index information.

[0118] The present invention realizes intelligent and humanized information push by constructing a powerful health information search engine and a TCM constitution test database and a dietary health database. From the aspects of user's personal data management, TCM constitution, common diseases, chronic disease information, health consultation, online inspection report viewing, dietary library, dietary mall platform, etc., by building a "health engine", targeted assistance and guidance are provided to users in terms of diet.

[0119] The following describes an embodiment of providing a healthy diet for users in combination with a number of specific examples.

[0120] In a specific example, when a user of a client searches and inputs the name of a food (ingredient), the information displayed to the client through the search engine of the management server is the nutritional content of the ingredient, suitable TCM constitution, suitable common diseases, suitable chronic diseases, suitable special populations, and suitable recipes, for the convenience of users.

[0121] In another specific example, when a user of a client searches and inputs the name of a recipe, the information displayed to the client through the search engine of the management server is the recipe preparation process, nutritional components of the food in the recipe, suitable TCM constitutions, suitable diseases, suitable special populations, etc., for the convenience of user use.

[0122] In another specific example, if a user of a client has set a disease, the management server will display to the client information suitable for the disease, the nutritional components of the food, suitable TCM constitution, suitable diseases, suitable special populations, corresponding recipes, etc. based on the user's disease type, for the convenience of the user.

[0123] In another specific example, after a user of a client is tested by a health information receiving unit, the user's TCM constitution can be obtained. Based on the user's TCM constitution, the system can recommend to the user foods suitable for this constitution, the nutritional components of the food, suitable special groups of people, corresponding recipes, etc.

[0124] The following describes an embodiment of providing personalized customized meal products to users in combination with a number of specific examples.

[0125] In a specific example, after a user of a client is tested by a health information receiving unit, the user's TCM constitution can be obtained. Based on the user's TCM constitution, the system can recommend dietary products suitable for this constitution to the user, and the user can purchase them directly through the dietary mall platform.

[0126] In another specific example, after the user connects his or her disease information or the subsequent hospital, the user's hospital visit data and the doctor's diagnosis data are connected to the system, and the system will provide guidance on healthy eating, thereby improving user satisfaction and stickiness to the system. The health engine can also gradually become more intelligent. For example, if the user enters text content such as "easy to get angry" in the communication area, the system will intelligently remind you of the diet plan to prevent getting angry.

[0127] Compared with the prior art, the advantages of the present invention are: by combining the dietary health database and the dietary mall platform with personal TCM constitution, chronic diseases, common diseases, and special populations, users are provided with tailor-made personalized dietary solutions. When the user's body is in a sub-healthy or diseased state, the client APP or mini-programs and other applications can be used to easily solve what to eat to avoid illness and what to eat to stay healthy, thereby avoiding the disease from the mouth, so as to improve the role of daily diet in promoting physical health and reduce the occurrence of diseases. Through personalized customized dietary solutions, users are helped to improve their dietary structure, balance their nutritional intake, and combine TCM theory to regulate and improve their physical condition, so as to achieve the purpose of healthy diet. By providing users with dietary intervention services anytime and anywhere, the health literacy of the whole people can be improved.

[0128] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned method for intelligently recommending health information based on big data when executing the computer program. The functions implemented by each functional module in this embodiment are the same as those of the above-mentioned system and method, and will not be described in detail here.

[0129] The present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the steps of the method for intelligent recommendation of health information based on big data as described above are implemented.

[0130] Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0131] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the present application can adopt the form of computer program products implemented on multiple computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that comprise computer-usable program codes.

[0132] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0133] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction device, which is implemented in the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0134] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0135] It should be noted that, in this application, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0136] The above description is only a specific implementation of the present application, so that those skilled in the art can understand or implement the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest range consistent with the principles and novel features applied for herein.

Claims

1. A health information intelligent recommendation system based on big data, characterized in that: Including client and management server: The client is used to receive a user's health information retrieval request and send the request to the management server through the network; The management server is provided with a TCM constitution test database, a health database including a food material database and a recipe database with TCM attributes, a dietary mall platform and a health information search engine. When receiving a health information search request sent by the client, the management server is used to control the health information search engine of the management server to perform a matching search in the health database and return the matching health data to the client through the network; In the process of processing the search request of the client, the management server first receives an initial search request from the client, retrieves the search results matching the initial search request from the diet health database, the disease database, and the traditional Chinese medicine constitution test database, and returns the search results to the client user; when a secondary search request indicating a selected search result in the search results is received from the client, the management server generates a matching link, the matching link is used to identify the hit database from which the selected search result is retrieved; the management server uses the matching link to link the client to the hit database identified in the matching link, so as to process the secondary search request in the hit database; After returning the matched health data to the client, the user's interest feedback information on the health data is introduced into the matching filtering algorithm, a set of similar users is returned to the user, and recommendations are made to the current user based on the interest feedback information of the similar user set; Assume that the number of interest feedback information of health recommendation results for users u and v is N, and the interest feedback information of user u is R u =(r u1 ,r u2 ,…,r uN ), user v’s interest feedback information Rv=(r v1 ,r v2 ,…,r vN ), r ui and r vi are the feedback information of user u and v on the i-th health recommendation result, i.e., the matching search item. The feedback information is expressed as satisfaction or interest level. The difference of feedback information of user u and v is expressed as: DELTA(u,v)=|R u -R v | The common interest feedback information difference DELTA (u, v) is dimensionally counted. Assuming there are n dimensions, its frequency distribution is expressed as: F(DELTA(u,v))=(p1,p2,…,p n ) Among them, p1, p2, …, p n represents the frequency of occurrence of each dimension, so the difference of F(DELTA(u,v)) is expressed as: H(F(DELTA(u,v)))=- In order to consider the difference of interest feedback information, the distribution probability is p i Interest feedback information difference |delta(p i )|Add to the above formula to get: sim H (u,v)= -×|delta(p i )| Calculate the weighted difference similarity sim WH (u,v): Yes WH (u,v)=yes H (u,v) f(u,v) f(u,v) is a weighted function, N is the number of common feedback information of users u and v, and f(u,v) is taken as: f(u,v)=1- Among them, the value range of α is 0<α<1; Personalized recommendations are made based on the interests of each user. Let the interest vector of user u be Ist u =(I u,1 ,I u,2 ,…,I u,n ) I u,i =c u,i / c u Among them, I u,i represents the interest vector of user u for the i-th recommendation result, c u,i is the number of recommended results for the i-th item whose interest level of user u is higher than the mean interest level, c u Represents the total number of all recommended results with a higher interest level than interest. The similarity calculation process of the interests of two users is as follows: simIst(u,v)=Jac(Ist u ,Is v ) Jac is the Jaccard similarity coefficient operation; through the similarity Sim(u,v) between users, the closest neighbor set v is determined for user u; Sim(u,v)=λ·sim WH (u,v)+(1−λ)·simIst(u,v) Where λ is the adjustment parameter; Based on the interest feedback information of similar user sets, optimized recommendations are made to the current user.

2. The health information intelligent recommendation system based on big data according to claim 1 is characterized in that: The dietary health database is based on the TCM constitution theory and basic TCM theory. It collects all common food ingredients, Chinese medicinal materials and food and medicine substances with the same origin as medicine according to the causes and symptoms of biased constitution, including food ingredient management and recipe management, wherein food ingredient management includes food ingredient type, TCM properties of neutral taste and meridians of food ingredients, main efficacy of food ingredients, nutritional components of food ingredients, and food taboos; the recipe management includes TCM efficacy of recipes, TCM properties of recipes, recipe cooking methods, and recipe taboos; wherein the taboos include analysis of the reasons for the taboos of food ingredients or recipes for TCM constitution, common diseases, chronic diseases, and special populations, and also includes the nutritional content of food ingredients and recipes suitable for different solar terms.

3. The health information intelligent recommendation system based on big data according to claim 1 is characterized in that: The dietary mall platform includes a personalized dietary mall platform, and the dietary mall platform includes a food library, food dimensions, food details, and a TCM constitution type suitable for the food.

4. The health information intelligent recommendation system based on big data according to claim 3 is characterized in that: The client is provided with a health information receiving unit, a disease information receiving unit and a special population setting unit; the health information receiving unit receives data input by the user, makes a judgment based on the data, and outputs the user's TCM constitution type; the disease information receiving unit allows the user to select the disease type, and the special population setting unit allows the user to select the special population, and the management server outputs health data suitable for the user to the client based on the user's TCM constitution type, disease type, and special population.

5. According to claim 1, the intelligent health information recommendation system based on big data is characterized in that: The management server receives the user's health risk data and user history data, processes the health risk data by using an unsupervised learning algorithm to generate a fixed-size vector representation, extracts health risk features from the vector representation by using a multi-layer convolutional neural network, determines dietary conditions associated with the health risk features, generates a dietary recommendation plan based on the dietary conditions and the user's history data, and displays the dietary recommendation plan for the user.

6. A health information intelligent recommendation method based on a health information intelligent recommendation system based on big data according to any one of claims 1 to 5, characterized in that: The following steps are involved: When the client receives a user's health information retrieval request, the client sends the request to the management server; The management server receives the user's health information search request and calls the health information search engine according to the user's request; The health information search engine performs engine data retrieval in the dietary health database; the health information search engine returns the retrieved matching data to the client.

7. The health information intelligent recommendation method according to claim 6, characterized in that: Also includes: The management server receives the user's health information retrieval request, and first normalizes the request according to a predefined data format, specifically dividing the request data format into two parts: a param part and a data part; the param part includes an MD5 signature for security verification; the data part is the input parameter data required for functional logic processing; the normalized data is input into the service platform entrance.

8. The health information intelligent recommendation method according to claim 7, characterized in that: Also includes: After the data is input into the service platform entrance, the parameters are converted and checked. The parameters include the service number serviceid, the data format dataformat, the user unique identifier sessionId, and the signature string MD5. The parameter conversion is performed through a converter, which converts the parameters from the " / / ; / / " and " / / = / / " formats into the framework flow object Param object; during the conversion process, the data is verified for validity, including format verification, service number verification, and MD5 signature verification, and then service routing is performed according to the service number serviceid.

9. The health information intelligent recommendation method according to claim 8, characterized in that: Also includes: The data after service routing is input into the pre-filter pool, which includes a decompression filter, a decryption filter, a remote service check filter, and a login filter. The filter pool also supports subsequent dynamic expansion. Only when all the checks of the pre-filter pool are passed, the retrieval service module of the health information search engine is called for logical processing.

10. The health information intelligent recommendation method according to claim 9, characterized in that: The health information search engine is used to perform keyword segmentation processing, and performs engine data retrieval in the health database based on the segmentation results. The health information search engine performs traversal queries, and during the query process, it will perform a matching score on the information entered by the user, thereby feeding back the valid search information set to the user end, and arranging the information in descending order according to the matching score; these matching data are then encapsulated, that is, the data is unified into the currently common JSON format.

11. The health information intelligent recommendation method according to claim 10, characterized in that: Also includes: The encapsulated JSON format data is input into a post-filter, which includes an encryption filter, a compression filter, and a result format encapsulation filter. The data output by the post-filter is routed via the service and then returned to the client via the service platform entrance.

12. A terminal device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method according to any one of claims 6 to 11 when executing the computer program.

13. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method according to any one of claims 6 to 11 are implemented.

Images