Health advice recommendation method, system and electronic device based on digital twin model

Abstract

The present application relates to the field of medical and health service technology, and discloses a health advice recommendation method, system and electronic device based on a digital twin model. The present application sets adaptive feature types for different advice strategies, and obtains the current feature type and current feature in the human digital twin model, so as to determine the target strategy from each advice strategy according to the matching result between the current feature type and the adaptive feature type, and executes the target strategy according to the current feature, so as to obtain the target health advice recommended based on the human digital twin model. Compared with the use of indiscriminate advice strategies for the human digital twin model, the application realizes the selection of advice strategies that conform to the human digital twin model through different feature types of the human digital twin model, thereby improving the accuracy of health advice recommendations when recommending health advice based on the human digital twin model.

Description

Technical Field

[0001] The present invention relates to the field of medical and health service technology, and in particular to a health advice recommendation method, system and electronic equipment based on a digital twin model. Background Art

[0002] In today's era, the rapid development of digital information technology is profoundly affecting the field of medical health. Among them, the integrated application of cutting-edge technologies such as big data, cloud computing and artificial intelligence has become a frontier hotspot for exploration in this field. Through the deep integration of "Internet +" and the medical and health industry, the health industry is radiating unprecedented vitality and vigor. Based on this, given that the human body is a highly complex and sophisticated system, the use of the human digital twin model to combine individual medical care information with huge health and medical big data can build a parallel interactive and mutually mapped system, thereby capturing and reflecting the individual's macro and micro, internal and external factors, physiological and psychological individualized characteristic data in an all-round and multi-dimensional manner, and realizing accurate digital representation of the human body state.

[0003] At present, the human body digital twin model is an advanced technology in the field of medical health in recent years. It is based on physical models, sensor updates, operation history and other data. It integrates multi-disciplinary, multi-physical quantity, multi-scale, and multi-probability simulation processes to create a "digital twin" corresponding to the real human body in the virtual space. This makes the human body digital twin model carry extremely rich information, and the health advice provided based on this model is therefore more accurate.

[0004] However, with the continuous evolution of human digital twin models and the increasing diversification of health advice strategies, the diversity of types and functions of the two is becoming increasingly significant. Among them, if only a single health advice strategy is adopted, it may face limitations such as incompatibility with specific types of human digital twin models and overly one-sided recommendations. If multiple human digital twin models are tried, it may cause problems such as high computing costs and contradictory recommendations. Therefore, there are generally problems of various types and insufficient compatibility between human digital twin models and health advice strategies, resulting in low accuracy of health advice recommendations based on human digital twin models. Summary of the invention

[0005] In order to provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. The summary is not an extensive review, nor is it intended to identify key / critical elements or delineate the scope of protection of these embodiments, but rather serves as a prelude to the detailed description that follows.

[0006] In view of the shortcomings of the prior art mentioned above, the present application provides a health advice recommendation method, system and electronic device based on a digital twin model, so as to improve the accuracy of health advice recommendations when recommending health advice based on a digital twin model of the human body.

[0007] The present application provides a health advice recommendation method based on a digital twin model, comprising: pre-setting multiple advice strategies, and setting the adaptation feature types corresponding to each of the advice strategies; obtaining a human digital twin model, wherein the human digital twin model includes one or more current feature types, and current features corresponding to each of the current feature types; determining a target strategy from each of the advice strategies according to the matching results between the current feature type and the adaptation feature type; executing the target strategy according to the current features to generate a target health advice corresponding to the human digital twin model.

[0008] In one embodiment of the present application, the target strategy is executed according to the current feature, including: if the target strategy includes a feature matching set, and the feature matching set is used to store multiple user health suggestions and model features corresponding to each of the user health suggestions, then the target health suggestion is determined from the user health suggestions according to the matching result between the current feature and the model feature; if the target strategy includes a health recognition model, and the health recognition model is obtained by model training based on sample features with health type labels, then the current feature is input into the health recognition model for recognition to obtain a health recognition result, and according to the health recognition result, matching is performed from a preset result matching set to obtain a target health suggestion, wherein the result matching set is used to store multiple user health suggestions and model output results corresponding to each of the user health suggestions; if the target strategy includes a language large model, the current feature is input into a preset prompt word template to obtain input data, and the input data is input into the language large model to obtain the target health suggestion output by the language large model, wherein the language large model is trained based on a preset knowledge pool containing the user health suggestion.

[0009] In one embodiment of the present application, a health recognition model is obtained in the following manner: a plurality of sample features and the health type corresponding to each of the sample features are obtained; a health type label corresponding to the sample feature is generated in a vector form according to the health type, wherein the health type label includes a probability distribution corresponding to each of the health types; the sample features with the health type label are used as training samples to perform model training on a preset neural network model, and the trained neural network model is determined as a health recognition model.

[0010] In one embodiment of the present application, the target strategy is executed according to the current feature to generate a target health recommendation corresponding to the human digital twin model, including: if the number of the target strategies is multiple, then the policy priority corresponding to each of the target strategies is obtained; according to the policy priority, the strategy to be executed is determined from each of the target strategies; in response to the strategy to be executed, the strategy to be executed is executed according to the current feature to obtain the execution result corresponding to the current feature, wherein the execution result includes the current health recommendation or the execution failure; if the execution result includes the execution failure, then a new strategy to be executed is determined from each of the target strategies according to the policy priority; the current health recommendations output by each of the target strategies are merged to generate the target health recommendation corresponding to the human digital twin model.

[0011] In one embodiment of the present application, an adaptive feature type corresponding to each of the suggested strategies is set, including: obtaining multiple original feature types, and obtaining feature samples corresponding to each of the original feature types; executing each of the suggested strategies according to the feature samples corresponding to different original feature types, and counting the strategy accuracy between each of the suggested strategies and each of the original feature types; if the strategy accuracy between the original feature type and the suggested strategy is greater than a preset accuracy threshold, the original feature type is determined as the adaptive feature type corresponding to the suggested strategy.

[0012] In one embodiment of the present application, a target strategy is determined from each of the recommended strategies based on the matching result between the current feature type and the adapted feature type, including: if the number of the current feature types is multiple, then matching is performed from each of the adapted feature types according to each of the current feature types respectively to obtain a target type, and the recommended strategy corresponding to the target type is determined as an alternative strategy; if the number of the alternative strategies is one, then the alternative strategy is determined as the target strategy; if the number of the alternative strategies is multiple, then calculation is performed based on the strategy accuracy between the target type and the alternative strategies to obtain a strategy adaptation score between the alternative strategies and the current feature type, and the target strategy is determined from the alternative strategies based on the strategy adaptation score, wherein there is a positive correlation between the strategy accuracy and the strategy adaptation score.

[0013] In one embodiment of the present application, before executing the target strategy according to the current feature, the method also includes: if the number of the current features is multiple, classifying the current features through a preset clustering algorithm to obtain multiple feature clusters; merging the current features in the same feature cluster to obtain merged features corresponding to the feature cluster; and updating the current features according to the merged features corresponding to each of the feature clusters.

[0014] In one embodiment of the present application, the current features in the same feature cluster are merged to obtain merged features corresponding to the feature cluster, including: performing text extraction from the current features in the feature cluster to obtain feature identifiers, and determining the health impact values ​​corresponding to each current feature in the feature cluster according to the health impact directions corresponding to each current feature in the feature cluster, wherein the health impact directions include positive correlation or negative correlation; calculating according to the impact relationships of each feature to obtain a merged feature value corresponding to the feature cluster; merging the feature identifiers and the merged feature values ​​to obtain a merged feature corresponding to the feature cluster.

[0015] The present application provides a health advice recommendation system based on a digital twin model, including: a setting module, used to set multiple advice strategies, and set the adaptation feature types corresponding to each of the advice strategies; an acquisition module, used to acquire a human digital twin model, wherein the human digital twin model includes one or more current feature types, and current features corresponding to each of the current feature types; a determination module, used to determine a target strategy from each of the advice strategies according to the matching results between the current feature type and the adaptation feature type; an execution module, used to execute the target strategy according to the current features to generate a target health advice corresponding to the human digital twin model.

[0016] The present application provides an electronic device, comprising: a processor and a memory; the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the electronic device executes the above method.

[0017] The present invention provides a computer-readable storage medium on which a computer program is stored: when the computer program is executed by a processor, the above method is implemented.

[0018] Beneficial effects of this application:

[0019] By setting the adaptation feature type for different recommendation strategies and obtaining the current feature type and current feature in the human digital twin model, the target strategy is determined from each recommendation strategy according to the matching result between the current feature type and the adaptation feature type, and the target strategy is executed according to the current feature to obtain the target health recommendation based on the human digital twin model. In this way, compared with the use of indiscriminate recommendation strategies for the human digital twin model, the human digital twin model is distinguished by the current feature type, so as to determine the target strategy from each recommendation strategy according to the matching result between the current feature type and the adaptation feature type, thereby realizing the selection of recommendation strategies that conform to the human digital twin model through different feature types of the human digital twin model, and then improving the accuracy of health recommendations when recommending health recommendations based on the human digital twin model. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a flowchart of a health advice recommendation method based on a digital twin model in an embodiment of the present application;

[0021] Figure 2 It is a flowchart of a method for determining a target strategy in an embodiment of the present application;

[0022] Figure 3 It is a schematic diagram of a DNN model training process in an embodiment of the present application;

[0023] Figure 4 It is a structural diagram of a health recognition model in an embodiment of the present application;

[0024] Figure 5 It is a flowchart of a target strategy execution method in an embodiment of the present application;

[0025] Figure 6 is a flowchart of a current feature updating method in an embodiment of the present application;

[0026] Figure 7 It is a flowchart of a current feature merging method in an embodiment of the present application;

[0027] Figure 8 It is a flowchart of another health advice recommendation method based on a digital twin model in an embodiment of the present application;

[0028] Fig. 9 It is a structural diagram of a health advice recommendation system based on a digital twin model in an embodiment of the present application;

[0029] Fig.10 It is a schematic diagram of the structure of an electronic device in an embodiment of the present invention. DETAILED DESCRIPTION

[0030] The following describes the embodiments of the present invention through specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and sub-samples in the embodiments can be combined with each other without conflict.

[0031] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and thus the drawings only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0032] In the following description, numerous details are discussed to provide a more thorough explanation of the embodiments of the present invention. However, it is obvious to those skilled in the art that the embodiments of the present invention can be implemented without these specific details. In other embodiments, well-known structures and devices are shown in the form of block diagrams rather than in detail to avoid making the embodiments of the present invention difficult to understand.

[0033] The terms "first", "second", etc. in the specification and claims of the embodiments of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the terms used in this way can be interchanged where appropriate, so that the embodiments of the embodiments of the present disclosure described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions.

[0034] Unless otherwise stated, the term "plurality" means two or more.

[0035] In the embodiment of the present disclosure, the character " / " indicates that the preceding and following objects are in an "or" relationship. For example, A / B indicates: A or B.

[0036] The term "and / or" is a description of the association relationship between objects, indicating that three relationships can exist. For example, A and / or B means: A or B, or, A and B.

[0037] Combination Figure 1 As shown, the embodiment of the present disclosure provides a health advice recommendation method based on a digital twin model, including:

[0038] Step S101, pre-setting a plurality of suggestion strategies, and setting the adaptation feature types corresponding to each suggestion strategy;

[0039] Step S102, obtaining a human body digital twin model;

[0040] The human digital twin model includes one or more current feature types and current features corresponding to each current feature type;

[0041] Step S103, determining a target strategy from among the suggested strategies according to the matching result between the current feature type and the adapted feature type;

[0042] Step S104, executing the target strategy according to the current features to generate target health recommendations corresponding to the human digital twin model.

[0043] The health advice recommendation method based on the digital twin model provided by the embodiment of the present disclosure is adopted. By setting the adaptation feature type for different advice strategies, and obtaining the current feature type and the current feature in the human digital twin model, the target strategy is determined from each advice strategy according to the matching result between the current feature type and the adaptation feature type, and the target strategy is executed according to the current feature to obtain the target health advice recommended based on the human digital twin model. In this way, compared with the use of indiscriminate advice strategies for the human digital twin model, the human digital twin model is distinguished by the current feature type, so as to determine the target strategy from each advice strategy according to the matching result between the current feature type and the adaptation feature type, thereby realizing the selection of advice strategies that conform to the human digital twin model through the different feature types of the human digital twin model, and then improving the recommendation accuracy of health advice when recommending health advice based on the human digital twin model.

[0044] Combination Figure 2 As shown, the embodiment of the present disclosure provides a method for determining a target strategy, including:

[0045] Step S201: setting the adaptation feature types corresponding to the respective suggested strategies.

[0046] Optionally, set the adaptation feature type corresponding to each recommended strategy, including:

[0047] Step S2011, obtaining a plurality of original feature types, and obtaining feature samples corresponding to each original feature type;

[0048] Step S2012, executing each recommended strategy according to the feature samples corresponding to different original feature types, and calculating the strategy accuracy between each recommended strategy and each original feature type;

[0049] Step S2013: If the strategy accuracy between the original feature type and the suggested strategy is greater than a preset accuracy threshold, the original feature type is determined as the adapted feature type corresponding to the suggested strategy.

[0050] In some embodiments, the original feature types include lifestyle, disease history, risk factors, physical examination indicators, dietary intake, etc.; if the current feature type is lifestyle, the current features corresponding to the lifestyle include smoke intake, secondhand smoke exposure duration, alcohol intake, meat intake, vegetable intake, exercise duration, sleep quality, etc.; if the current feature type is disease history, the current features corresponding to the disease history include own hypertension, own ventilation, own gastroesophageal reflux, maternal hypertension, grandfather's lung cancer, etc.; if the current feature type is risk factors, the current features corresponding to the risk factors include excessive smoke intake, excessive secondhand smoke exposure duration, excessive alcohol intake, insufficient vegetable input, insufficient grain intake, etc.; if the current feature type is physical examination indicators, the current features corresponding to the physical examination indicators include height, weight, vital capacity, blood pressure value, organ imaging, etc.; if the current feature type is dietary intake, the current features corresponding to the dietary intake include meat intake, vegetable intake, grain intake, calcium intake, vitamin intake, drug intake, etc.

[0051] In some embodiments, the multiple recommendation strategies include at least two of a feature matching set, a health recognition model, and a large language model, wherein the feature matching set focuses more on the feature itself and is suitable for independent feature types such as physical examination indicators and risk factors; the health recognition model processes scenarios where features are more dispersed, such as feature types such as lifestyle; the large language model can provide effective health advice when processing a wide range of contextual information, and is suitable for scenarios where there are extensive connections between features such as lifestyle and disease history.

[0052] Step S202: determining a target strategy from among the suggested strategies according to the matching result between the current feature type and the adapted feature type.

[0053] Optionally, determining a target strategy from each of the suggested strategies according to a matching result between the current feature type and the adapted feature type includes:

[0054] Step S2021: if there are multiple current feature types, then match each current feature type from each adapted feature type to obtain a target type, and determine the recommended strategy corresponding to the target type as an alternative strategy;

[0055] Step S2022: if the number of candidate strategies is one, determine the candidate strategy as the target strategy;

[0056] Step S2023, if there are multiple candidate strategies, calculate the strategy accuracy between the target type and the candidate strategies to obtain the strategy adaptation score between the candidate strategies and the current feature type, and determine the target strategy from the candidate strategies according to the strategy adaptation score;

[0057] Among them, there is a positive correlation between strategy accuracy and strategy adaptation score.

[0058] In some embodiments, the current feature type includes type a and type b, and the alternative strategies include strategy A, strategy B and strategy C, wherein the target type corresponding to strategy A is type a, the target type corresponding to strategy B is type a and type b, and the target type corresponding to strategy C is type a and type b; the strategy adaptation value is determined according to the strategy accuracy between the target type and the alternative strategy, wherein the strategy accuracy and the strategy adaptation value are nonlinearly positively correlated, that is, the rate of change of the strategy adaptation value is greater than the rate of change of the strategy accuracy, so that the recommended strategy with high strategy accuracy has a higher weight. For example, if the strategy accuracy between type a and strategy B is 90%, the strategy adaptation value between type a and strategy B is 0.9, if the strategy accuracy between type b and strategy B is 50%, then the strategy adaptation value between type b and strategy B is 0.2, and the strategy accuracy is set as parameter , set the policy adaptation value to parameter , set up the nonlinear equation ,according to and Fit or interpolate nonlinear equations to obtain coefficients ,coefficient , through the obtained equation Calculate the policy adaptation value corresponding to the policy accuracy; determine the policy adaptation score between the current feature type and the alternative strategy through the policy adaptation value. For example, the policy adaptation score of strategy B and the current feature type is 0.9+0.2=1.1; when the policy adaptation score is greater than the preset score threshold, determine the alternative strategy corresponding to the policy adaptation score as the target strategy, thereby achieving feature type differentiation and making the health recommendations based on the current features more accurate.

[0059] Optionally, executing a target strategy according to the current feature includes: if the target strategy includes a feature matching set, and the feature matching set is used to store multiple user health suggestions and model features corresponding to each user health suggestion, then determining the target health suggestion from the user health suggestions based on the matching results between the current feature and the model feature; if the target strategy includes a health recognition model, and the health recognition model is obtained by model training based on sample features with health type labels, then the current feature is input into the health recognition model for recognition to obtain a health recognition result, and matching is performed from a preset result matching set based on the health recognition result to obtain the target health suggestion, wherein the result matching set is used to store multiple user health suggestions and model output results corresponding to each user health suggestion; if the target strategy includes a large language model, then the current feature is input into a preset prompt word template to obtain input data, and the input data is input into the large language model to obtain the target health suggestion output by the large language model, wherein the large language model is trained based on a preset knowledge pool containing user health suggestions.

[0060] In some embodiments, the similarity between the current feature and the model feature is calculated through a similarity algorithm to determine the model feature matched by the current feature based on the calculated similarity. For example, the feature matching set includes the model feature matched by the current feature "alcohol intake"; the corresponding user health advice is extracted based on the model feature matched by the current feature "alcohol intake" through the feature matching set.

[0061] In some embodiments, similarity algorithms include cosine similarity, Jaccard similarity, Euclidean distance, Manhattan distance, Levenshtein distance, semantic similarity, sentence embedding, etc.

[0062] Optionally, the health recognition model is obtained in the following manner: obtaining multiple sample features and the health type corresponding to each sample feature; generating a health type label corresponding to the sample feature in a vector form according to the health type, wherein the health type label includes the probability distribution corresponding to each health type; using the sample features with the health type label as training samples to perform model training on a preset neural network model, and determining the trained neural network model as the health recognition model.

[0063] In some embodiments, multiple health types and corresponding identification bits for each health type are predefined; in order to clearly represent different health types, health type labels representing the health types are generated in vector form, wherein different identification bits in the vector identify different health types, and the corresponding probability distributions are probability values ​​corresponding to different health types, respectively; during the training process of the health recognition model, sample features with health type labels are input into the health recognition model as training samples to obtain output data with probability distribution, and the probability value of the output data can be used to determine from the health types which health type the human digital twin model belongs to.

[0064] In some embodiments, the neural network model includes a deep neural network (DNN) model, wherein the DNN model consists of multiple hidden layers and has powerful representation capabilities for processing various types of medical data, including numerical data (such as laboratory test results), categorical data (such as health diagnosis), etc.; the DNN model can automatically learn complex patterns and relationships in the data through multiple layers of nonlinear transformations.

[0065] Combination Figure 3 As shown, the present disclosure provides a DNN model training method, including:

[0066] Step S301, data preprocessing, including cleaning, standardization, encoding and other processing of training samples to make them suitable for the input of the DNN model;

[0067] Step S302, constructing a model, including determining the number of layers of the DNN model, the number of neurons in each layer, the activation function and other parameters;

[0068] Step S303, training the model, including training the DNN model using the preprocessed medical data, so as to continuously adjust the parameters of the model through an optimization algorithm, so that the model can minimize the loss function and accurately identify the health type of the current user;

[0069] Step S304, recommendation generation, inputs the user's medical features into the trained DNN model, the model outputs the prediction results, and generates health recognition results based on the prediction results.

[0070] Combination Figure 4As shown, the health recognition model obtained by training the DNN model includes an input layer, a hidden layer and an output layer; the input layer is the layer where the deep neural network receives input data, which is used to convert each word into a corresponding word vector representation so that the subsequent neural network layer can process it, wherein the current data includes the current features; the hidden layer is the core part of the deep neural network, which is composed of multiple neurons. The function of the hidden layer is to perform nonlinear transformation on the input data, extract features from the data, and pass these features to the next layer, wherein the deep neural network is usually composed of multiple hidden layers. The more layers there are, the stronger the representation ability of the network is. The number of neurons in each hidden layer will also affect the performance of the network. Generally speaking, the more neurons there are, the richer the features that the network can learn, but it will also increase the amount of calculation and the risk of overfitting; the output layer is the last layer of the deep neural network, and its function is to convert the features extracted by the hidden layer into the final output result. In medical knowledge recommendation, the output of the output layer can be recommended medical knowledge, disease diagnosis results, etc. The number of neurons and the selection of activation function of the output layer depend on the specific task requirements. For example, if it is a binary classification task (such as the presence or absence of disease), the output layer can have only one neuron, using the sigmoid activation function, and the output value is between 0 and 1, indicating the probability of disease occurrence. If a multi-classification task is being performed (such as the type of disease), the number of neurons in the output layer can be equal to the number of categories, and the softmax activation function is used, and the output value represents the probability distribution of each category.

[0071] In some embodiments, neurons are the basic components of the hidden layer. Each neuron receives an input signal from the previous layer and generates an output signal through weighted summation and activation function processing. For example, a neuron can receive multiple input signals x1, x2, ..., xn, each of which has a corresponding weight w1, w2, ..., wn. The neuron performs weighted summation on these input signals, that is, , b is the bias term, and the weighted summation result z is input into the activation function f(z) to obtain the output signal y=f(z) of the neuron.

[0072] In some embodiments, the role of the activation function is to introduce nonlinear characteristics to the neural network so that the network can learn complex functional relationships. Common activation functions include sigmoid function, tanh function, ReLU function, etc.; for example, when the input of the ReLU function (Rectified Linear Unit) is positive, the output is equal to the input; when the input is negative, the output is 0. This characteristic enables the ReLU function to effectively alleviate the gradient vanishing problem when training deep neural networks and improve training efficiency.

[0073] In some embodiments, connection weights and bias terms are parameters in a deep neural network that are used to adjust the connection strength between neurons and the output of neurons; during the training process, the connection weights and bias terms are continuously adjusted through an optimization algorithm so that the output of the network is as close to the true label value as possible; the connection weights represent the degree of influence of the output of the neurons in the previous layer on the neurons in the next layer; the bias term is a constant that is used to adjust the output of neurons; the initial values ​​of the connection weights and bias terms are usually randomly generated and then updated through a back-propagation algorithm during the training process.

[0074] In some embodiments, the role of the optimization algorithm is to adjust the connection weights and bias terms during the training process to minimize the loss function of the network; common optimization algorithms include stochastic gradient descent (SGD), Adam optimization algorithm, etc.; for example, the stochastic gradient descent algorithm calculates the gradient of the loss function with respect to the connection weights and bias terms, and then updates the parameters in the opposite direction of the gradient to gradually reduce the value of the loss function.

[0075] In some embodiments, the loss function is used to measure the difference between the network output and the true label value; in medical knowledge recommendation, the loss function can be selected according to the specific task requirements; common loss functions include mean squared error (MSE), cross entropy loss function, etc.; for example, in a binary classification task, the cross entropy loss function can be used, and its calculation formula is ,in, is the true label value (0 or 1), and p is the probability value predicted by the network; the smaller the value of the cross entropy loss function, the closer the network's prediction result is to the true label value.

[0076] In some embodiments, the health identification results output by the health identification model are matched with the result matching set to obtain targeted health recommendations. For example, if the health identification result is diabetes, the result matching set can recommend diet recommendations, exercise plans, drug treatment knowledge, etc. suitable for the patient.

[0077] In some embodiments, a knowledge pool is constructed based on medical common sense and expert knowledge, and a language model is trained through the knowledge pool so that the language model can recommend appropriate target health advice based on current features. For example, if the current user is a diabetic patient, the language model recommends diet advice, exercise plans, drug treatment knowledge, etc. suitable for the patient based on factors such as blood sugar control and complication risks in the current features. At the same time, experts also recommend the latest research results, treatment guidelines and case analyses based on their professional fields and clinical experience to improve the accuracy and efficiency of health advice.

[0078] Combination Figure 5 As shown, the embodiment of the present disclosure provides a target policy execution method based on step S104, including:

[0079] Step S1041, if there are multiple target policies, obtain the policy priority corresponding to each target policy;

[0080] Step S1042, determining the policy to be executed from each target policy according to the policy priority;

[0081] Step S1043, in response to the strategy to be executed, executing the strategy to be executed according to the current feature, and obtaining an execution result corresponding to the current feature;

[0082] Among them, the execution result includes current health advice and / or execution failure;

[0083] Step S1044, if the execution result includes execution failure, then determining a new strategy to be executed from each target strategy according to the strategy priority;

[0084] Step S1045, the current health recommendations output by each target strategy are integrated to generate target health recommendations corresponding to the human digital twin model.

[0085] In some embodiments, if the target policy includes a feature matching set and a health recognition model, and the priority of the feature matching set is higher than that of the health recognition model, the policy to be executed is the feature matching set. Therefore, the current feature is first matched through the feature matching set; the recognition result corresponding to the current feature is determined, and if the execution result includes the current health advice, the current health advice is the health advice output by the feature matching set; if there is a current feature that fails to match, that is, the execution result of the current feature is an execution failure, then according to the policy priority, the new policy to be executed is the health recognition model, and the current feature that fails to match is used as the feature to be executed, and is input into the health recognition model for identification, and the current health advice corresponding to the current feature corresponding to the execution result of the execution failure is obtained, that is, the health advice output by the input health recognition model; the health advice output by the feature matching set and the health recognition model are merged to obtain the target health advice.

[0086] It is not difficult to understand that if the target strategy includes a feature matching set and a large language model, and the priority of the feature matching set is higher than that of the large language model, the strategy to be executed is the feature matching set, that is, the feature matching set is executed first. If the current feature identified by the feature matching set fails to execute, the large language model is used to output health recommendations for the current feature that failed to execute. The specific execution process is similar to the above content and will not be repeated here.

[0087] It is not difficult to understand that if the target strategy includes feature matching set, health recognition model and language large model, and the strategy priorities are arranged from high to low: feature matching set, health recognition model, language large model. That is, the feature matching set is executed first. If the current feature executed by the feature matching set fails to be recognized, the health recognition model is used to perform recognition results on the current feature that failed to be executed. If the current feature executed by the health recognition model set fails to be recognized, the language large model is used to perform recognition results on the current feature that failed to be executed. The specific execution process is similar to the above content and will not be repeated here.

[0088] Combination Figure 6 As shown, before executing the target strategy according to the current feature, the embodiment of the present disclosure provides a current feature update method, including:

[0089] Step S601, if there are multiple features, classify the current features using a preset clustering algorithm to obtain multiple feature clusters;

[0090] Step S602, merging the current features in the same feature cluster to obtain a merged feature corresponding to the feature cluster;

[0091] Step S603: update the current feature according to the merged features corresponding to each feature cluster.

[0092] In some embodiments, the preset clustering algorithm includes one or more of a K-means clustering algorithm, a hierarchical clustering algorithm, a DBSCAN density clustering algorithm, and the like.

[0093] Combination Figure 7 As shown, the embodiment of the present disclosure provides a current feature merging method based on step S602, including:

[0094] Step S6021, performing text extraction from the current feature in the feature cluster to obtain a feature identifier, and determining the health impact value corresponding to each current feature in the feature cluster according to the health impact direction corresponding to each current feature in the feature cluster;

[0095] Among them, the direction of health impact includes positive or negative correlation;

[0096] Step S6022, calculating according to the influence relationship of each feature to obtain a combined feature value corresponding to the feature cluster;

[0097] Step S6023, merging is performed according to the feature identifier and the merged feature value to obtain a merged feature corresponding to the feature cluster.

[0098] In some embodiments, the health impact direction and health impact value of a feature are set by an expert.

[0099] In some embodiments, if the current features in the same feature cluster include "presence of smoke in the occupational environment", "smoke intake", and "duration of exposure to secondhand smoke", and each current feature has a negative correlation with physical health, then the health impact value set for each current feature is a negative value, and the merged feature value calculated based on the health impact value is also a negative value, for example, the merged feature value is -5; the feature identifier of the feature cluster is determined to be "lungs", and the merged feature value is merged with the feature identifier to obtain "lungs -5 level".

[0100] In some embodiments, if the current features in the same feature cluster include "higher sleep quality" and "sleep time less than 8 hours", higher sleep quality has a positive correlation with physical health, while "sleep time less than 8 hours" has a negative correlation with physical health; the merged feature value calculated according to the health impact value is 0; the feature identifier of the feature cluster is determined as "sleep", and the merged feature value is merged with the feature identifier to obtain "sleep level 0".

[0101] Combination Figure 8 As shown, the embodiment of the present disclosure provides a health advice recommendation method based on a digital twin model, including:

[0102] Step S801, setting multiple suggestion strategies;

[0103] Among them, the suggested strategies include feature matching sets, health recognition models, and language large models;

[0104] Step S802, setting the adaptation feature type corresponding to each suggested strategy according to the strategy accuracy rate corresponding to the adaptation feature type of the suggested strategy;

[0105] Step S803, setting a strategy adaptation value according to the strategy accuracy rate corresponding to the adaptation feature type of the recommended strategy;

[0106] Step S804, obtaining a human body digital twin model;

[0107] The human digital twin model includes one or more current feature types and current features corresponding to each current feature type;

[0108] Step S805, calculating according to the strategy adaptation value to obtain the strategy adaptation scores corresponding to the current feature type in different recommended strategies, so as to determine the target strategy from the recommended strategies;

[0109] Step S806, executing the target strategy according to the current characteristics, and obtaining health suggestions recommended by each target strategy;

[0110] Step S807, generating target health recommendations corresponding to the human digital twin model according to the health recommendations recommended by each target strategy.

[0111] The health advice recommendation method based on the digital twin model provided by the embodiment of the present disclosure is adopted. By setting the adaptation feature type for different advice strategies, and obtaining the current feature type and the current feature in the human digital twin model, the target strategy is determined from each advice strategy according to the matching result between the current feature type and the adaptation feature type, and the target strategy is executed according to the current feature to obtain the target health advice recommended based on the human digital twin model. In this way, compared with the use of indiscriminate advice strategies for the human digital twin model, the human digital twin model is distinguished by the current feature type, so as to determine the target strategy from each advice strategy according to the matching result between the current feature type and the adaptation feature type, thereby realizing the selection of advice strategies that conform to the human digital twin model through the different feature types of the human digital twin model, and then improving the recommendation accuracy of health advice when recommending health advice based on the human digital twin model.

[0112] Combination Fig. 9 As shown, an embodiment of the present disclosure provides a health advice recommendation system based on a digital twin model, including a setting module 901, an acquisition module 902, a determination module 903 and an execution module 904.

[0113] The setting module 901 is used to set a plurality of suggestion strategies and set the adaptation feature type corresponding to each suggestion strategy.

[0114] The acquisition module 902 is used to acquire a human digital twin model, wherein the human digital twin model includes one or more current feature types and current features corresponding to each current feature type.

[0115] The determination module 903 is used to determine the target strategy from the suggested strategies according to the matching result between the current feature type and the adapted feature type.

[0116] The execution module 904 is used to execute the target strategy according to the current characteristics to generate target health recommendations corresponding to the human digital twin model.

[0117] The health advice recommendation system based on the digital twin model provided by the embodiment of the present disclosure is adopted. By setting the adaptation feature type for different advice strategies, and obtaining the current feature type and the current feature in the human digital twin model, the target strategy is determined from each advice strategy according to the matching result between the current feature type and the adaptation feature type, and the target strategy is executed according to the current feature to obtain the target health advice recommended based on the human digital twin model. In this way, compared with the use of indiscriminate advice strategies for the human digital twin model, the human digital twin model is distinguished by the current feature type, so as to determine the target strategy from each advice strategy according to the matching result between the current feature type and the adaptation feature type, thereby realizing the selection of advice strategies that conform to the human digital twin model through the different feature types of the human digital twin model, and then improving the recommendation accuracy of health advice when recommending health advice based on the human digital twin model.

[0118] An embodiment of the present disclosure further provides an electronic device, including: a processor and a memory; the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the electronic device executes the above method.

[0119] Fig.10 The structure diagram of the computer system suitable for implementing the electronic device of the embodiment of the present application is shown. It should be noted that: Fig.10 The computer system 1000 of the electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0120] like Fig.10 As shown, the computer system 1000 includes a central processing unit (CPU) 1001, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 1002 or the program loaded from the storage part 1008 to the random access memory (RAM) 1003, such as executing the method in the above embodiment. Various programs and data required for system operation are also stored in the RAM 1003. The CPU 1001, the ROM 1002 and the RAM 1003 are connected to each other through the bus 1004. The input / output (I / O) interface 1005 is also connected to the bus 1004.

[0121] The following components are connected to the I / O interface 1005: an input section 1006 including a keyboard, a mouse, etc.; an output section 1007 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 1008 including a hard disk, etc.; and a communication section 1009 including a network interface card such as a LAN (Local Area Network) card, a modem, etc. The communication section 1009 performs communication processing via a network such as the Internet. A drive 1010 is also connected to the I / O interface 1005 as needed. A removable medium 1011, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1010 as needed so that a computer program read therefrom is installed into the storage section 1008 as needed.

[0122] The electronic device disclosed in this embodiment includes a processor, a memory, a transceiver and a communication interface. The memory and the communication interface are connected to the processor and the transceiver and complete communication with each other. The memory is used to store computer programs, the communication interface is used to communicate, and the processor and the transceiver are used to run the computer program so that the electronic device executes each step of the above method.

[0123] The embodiments of the present disclosure further provide a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, any one of the methods in the embodiments is implemented.

[0124] The computer-readable storage medium in the embodiments of the present disclosure can be understood by ordinary technicians in the field: all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to the computer program. The aforementioned computer program can be stored in a computer-readable storage medium. When the program is executed, the execution includes the steps of the above-mentioned method embodiments; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk and other media that can store program codes.

[0125] The above description and the accompanying drawings fully illustrate the embodiments of the present disclosure so that those skilled in the art can practice them. Other embodiments may include structural, logical, electrical, process and other changes. The embodiments represent only possible changes. Unless explicitly required, separate components and functions are optional, and the order of operations may vary. Parts and sub-samples of some embodiments may be included in or replace parts and sub-samples of other embodiments. Moreover, the words used in this application are only used to describe the embodiments and are not used to limit the claims. As used in the description of the embodiments and the claims, unless the context clearly indicates, the singular forms of "a", "an" and "the" are intended to include plural forms as well. Similarly, the term "and / or" as used in this application refers to any and all possible combinations of one or more associated listings. In addition, when used in this application, the term "comprise" and its variants "comprises" and / or comprising refer to the presence of stated sub-samples, wholes, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other sub-samples, wholes, steps, operations, elements, components and / or groups thereof. In the absence of further restrictions, the elements defined by the sentence "including one..." do not exclude the presence of other identical elements in the process, method or device including the elements. In this article, each embodiment may focus on the differences from other embodiments, and the same and similar parts between the embodiments may refer to each other. For the methods, products, etc. disclosed in the embodiments, if they correspond to the method part disclosed in the embodiments, then the relevant parts can refer to the description of the method part.

[0126] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software may depend on the specific application and design constraints of the technical solution. Technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the embodiments of the present disclosure. Technicians can clearly understand that for the convenience and simplicity of description, the specific working process of the above-described system, device and unit can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0127] In the embodiments disclosed herein, the disclosed methods and products (including but not limited to devices, equipment, etc.) can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of units can be only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some sub-samples can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between each other shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms. The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to implement this embodiment. In addition, each functional unit in the embodiment of the present disclosure may be integrated in a processing unit, or each unit may exist physically alone, or two or more units may be integrated in one unit.

[0128] The flowchart and block diagram in the accompanying drawings show the possible architecture, functions and operations of the system, method and computer program product according to the embodiment of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of the code, and a part of the module, program segment or code contains one or more executable instructions for realizing the specified logical function. In some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. In the description corresponding to the flowchart and block diagram in the accompanying drawings, the operations or steps corresponding to different boxes can also occur in a different order from the order disclosed in the description, and sometimes there is no specific order between different operations or steps. For example, two consecutive operations or steps can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, which can depend on the functions involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented by a dedicated hardware-based system that performs the specified functions or actions, or may be implemented by a combination of dedicated hardware and computer instructions.

Claims

1. A health advice recommendation method based on a digital twin model, characterized in that: include: Pre-setting a plurality of suggestion strategies and setting the adaptation feature types corresponding to each of the suggestion strategies; Acquire a human digital twin model, wherein the human digital twin model includes one or more current feature types and current features corresponding to each of the current feature types; Determining a target strategy from each of the suggested strategies according to a matching result between the current feature type and the adapted feature type; Executing the target strategy according to the current feature to generate a target health recommendation corresponding to the human digital twin model; Executing the target strategy according to the current feature, including: if the target strategy includes a feature matching set, and the feature matching set is used to store multiple user health suggestions and model features corresponding to each of the user health suggestions, then determining the target health suggestion from the user health suggestions according to the matching result between the current feature and the model feature; if the target strategy includes a health recognition model, and the health recognition model is obtained by model training based on sample features with health type labels, then the current feature is input into the health recognition model for recognition to obtain a health recognition result, and matching is performed from a preset result matching set based on the health recognition result to obtain a target health suggestion, wherein the result matching set is used to store multiple user health suggestions and model output results corresponding to each of the user health suggestions; if the target strategy includes a language macro model, then the current feature is input into a preset prompt word template to obtain input data, and the input data is input into the language macro model to obtain the target health suggestion output by the language macro model, wherein the language macro model is trained based on a preset knowledge pool containing the user health suggestions.

2. The method according to claim 1, characterized in that Get the health recognition model by: Acquire multiple sample characteristics and health types corresponding to the sample characteristics; Generating a health type label corresponding to the sample feature in a vector form according to the health type, wherein the health type label includes a probability distribution corresponding to each of the health types; The sample features with health type labels are used as training samples to train the preset neural network model, and the trained neural network model is determined as a health recognition model.

3. The method according to claim 1, characterized in that Executing the target strategy according to the current feature to generate a target health recommendation corresponding to the human digital twin model includes: If there are multiple target policies, obtain the policy priorities corresponding to each target policy; Determining a strategy to be executed from each of the target strategies according to the strategy priority; In response to the strategy to be executed, executing the strategy to be executed according to the current feature, and obtaining an execution result corresponding to the current feature, wherein the execution result includes current health advice and / or execution failure; If the execution result includes execution failure, determining a new strategy to be executed from each of the target strategies according to the strategy priority; The current health recommendations output by each of the target strategies are integrated to generate target health recommendations corresponding to the human digital twin model.

4. The method according to any one of claims 1 to 3, characterized in that: The adaptation feature types corresponding to the suggested strategies are set, including: Acquire multiple original feature types, and acquire feature samples corresponding to each of the original feature types; Execute each of the suggested strategies according to the feature samples corresponding to different original feature types, and count the strategy accuracy between each of the suggested strategies and each of the original feature types; If the strategy accuracy between the original feature type and the suggested strategy is greater than a preset accuracy threshold, the original feature type is determined as the adapted feature type corresponding to the suggested strategy.

5. The method according to claim 4, characterized in that Determining a target strategy from each of the suggested strategies according to a matching result between the current feature type and the adapted feature type includes: If there are multiple current feature types, matching is performed from each of the adapted feature types according to each of the current feature types to obtain a target type, and a recommended strategy corresponding to the target type is determined as an alternative strategy; If the number of the candidate strategies is one, determining the candidate strategy as the target strategy; If there are multiple alternative strategies, the strategy accuracy between the target type and the alternative strategies is calculated to obtain the strategy adaptation score between the alternative strategies and the current feature type, and the target strategy is determined from the alternative strategies based on the strategy adaptation score, wherein there is a positive correlation between the strategy accuracy and the strategy adaptation score.

6. The method according to any one of claims 1 to 3, characterized in that: Before executing the target strategy according to the current feature, the method further includes: If there are multiple current features, the current features are classified by a preset clustering algorithm to obtain multiple feature clusters; Merging the current features in the same feature cluster to obtain merged features corresponding to the feature cluster; The current feature is updated according to the merged features respectively corresponding to the feature clusters.

7. The method according to claim 6, characterized in that Merging the current features in the same feature cluster to obtain the merged features corresponding to the feature cluster includes: Performing text extraction from the current feature in the feature cluster to obtain a feature identifier, and determining the health impact value corresponding to each current feature in the feature cluster according to the health impact direction corresponding to each current feature in the feature cluster, wherein the health impact direction includes positive correlation or negative correlation; Calculate according to each of the feature influence relationships to obtain a combined feature value corresponding to the feature cluster; The feature identifier and the merged feature value are merged to obtain a merged feature corresponding to the feature cluster.

8. A health advice recommendation system based on a digital twin model, characterized in that: include: A setting module, used to set a plurality of suggestion strategies and set the adaptation feature type corresponding to each of the suggestion strategies; An acquisition module, used to acquire a human digital twin model, wherein the human digital twin model includes one or more current feature types and current features corresponding to each of the current feature types; A determination module, configured to determine a target strategy from among the suggested strategies according to a matching result between the current feature type and the adapted feature type; An execution module, configured to execute the target strategy according to the current feature to generate a target health recommendation corresponding to the human digital twin model; The execution module executes the target strategy according to the current feature in the following manner: if the target strategy includes a feature matching set, and the feature matching set is used to store multiple user health suggestions and model features corresponding to each of the user health suggestions, then the target health suggestion is determined from the user health suggestions according to the matching result between the current feature and the model feature; if the target strategy includes a health recognition model, and the health recognition model is obtained by model training based on sample features with health type labels, then the current feature is input into the health recognition model for recognition to obtain a health recognition result, and the health recognition result is matched from a preset result matching set to obtain a target health suggestion, wherein the result matching set is used to store multiple user health suggestions and model output results corresponding to each of the user health suggestions; if the target strategy includes a language macro model, then the current feature is input into a preset prompt word template to obtain input data, and the input data is input into the language macro model to obtain the target health suggestion output by the language macro model, wherein the language macro model is trained based on a preset knowledge pool containing the user health suggestions.

9. An electronic device, characterized in that: include: Processor and memory; The memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the electronic device executes the method according to any one of claims 1 to 7.

Images