A physiological data analysis method and device, electronic equipment and storage medium

Abstract

The application relates to the technical field of data analysis, in particular to the field of artificial intelligence, and provides a physiological data analysis method and device, an electronic device and a storage medium, to improve the analysis accuracy of physiological data. The method comprises the following steps: obtaining original state information and at least one type of object attribute information of an object to be analyzed; performing feature extraction on the original state information based on the conditional probability configured for each preset state information, to obtain a corresponding first intermediate result; performing feature extraction on the at least one type of object attribute information respectively, to obtain a corresponding second intermediate result of each type of object attribute information; determining the category probability of each type of preset physiological data analysis result corresponding to the object to be analyzed based on the first intermediate result and the at least one second intermediate result, and determining the target physiological data analysis result of the object to be analyzed based on each category probability. In this way, the analysis accuracy can be effectively improved by combining the conditional probability configured for the corresponding preset state information.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data analysis, in particular to the technical field of artificial intelligence, and provides a physiological data analysis method and device, an electronic device and a storage medium. BACKGROUND

[0002] With the development of computer technology, computers are increasingly used in the medical field, for example, electronic physiological data of a subject is collected for medical statistical analysis, and neural networks are used to analyze physiological data of a subject.

[0003] For example, in intelligent diagnosis, a subject can input physiological data such as symptoms to an automatic diagnosis system, and obtain analysis results that may match the symptoms output by the automatic diagnosis system, so that the subject can make follow-up medical treatment or self-treatment according to the obtained analysis results.

[0004] In the related art automatic diagnosis system, the physiological data analysis method is generally based on a large amount of labeled data to establish a machine learning model or a neural network model, and a large amount of data is required to establish the machine learning model or the neural network model, and the labeling cost is high. Using a knowledge-driven method, the knowledge cannot be automatically adjusted according to the data, a large amount of manpower is required to manually adjust and correct the knowledge, the results after adjusting each piece of knowledge cannot be viewed in real time, and misjudgment is likely to occur, which affects the accuracy.

[0005] Therefore, how to effectively improve the analysis accuracy of physiological data is an urgent problem to be solved. SUMMARY

[0006] The embodiments of the present application provide a physiological data analysis method, device, electronic device and storage medium to improve the analysis accuracy of physiological data.

[0007] The physiological data analysis method provided by the embodiments of the present application comprises:

[0008] Obtaining original state information of a subject to be analyzed and at least one type of object attribute information related to the subject to be analyzed; wherein the original state information comprises physiological data of the subject to be analyzed in an environment in which the subject to be analyzed is located;

[0009] Based on a conditional probability configured for each preset state information, feature extraction is performed on the original state information to obtain a corresponding first intermediate result; the conditional probability represents a probability of occurrence of a preset state information under a condition of a preset physiological data analysis result;

[0010] Respectively performing feature extraction on the at least one type of object attribute information to obtain a second intermediate result corresponding to each of the at least one type of object attribute information;

[0011] determine the target physiological data analysis result of the to-be-analyzed object based on the category probabilities.

[0012] An embodiment of the present application provides a physiological data analysis device, which comprises:

[0013] an information acquisition unit, configured to acquire original state information of a to-be-analyzed object and at least one type of object attribute information related to the to-be-analyzed object, wherein the original state information comprises physiological data of the to-be-analyzed object in an environment in which the to-be-analyzed object is located;

[0014] a feature extraction unit, configured to perform feature extraction on the original state information based on conditional probabilities corresponding to each preset state information configuration, to obtain a corresponding first intermediate result; the conditional probability represents a probability of occurrence of a preset state information under a preset physiological data analysis result;

[0015] The feature extraction unit is further configured to perform feature extraction on the at least one type of object attribute information respectively, to obtain a corresponding second intermediate result of each type of object attribute information.

[0016] a result analysis unit, configured to determine category probabilities of each type of preset physiological data analysis result corresponding to the to-be-analyzed object based on the first intermediate result and at least one second intermediate result, and determine a target physiological data analysis result of the to-be-analyzed object based on the category probabilities.

[0017] Optionally, the feature extraction unit is specifically configured to:

[0018] acquire the conditional probability of each preset state information under each preset physiological data analysis result respectively, and determine a first weight matrix used for feature extraction based on the acquired conditional probabilities;

[0019] perform feature extraction on the original state information based on the first weight matrix, to obtain a corresponding first intermediate result.

[0020] Optionally, the feature extraction unit is specifically configured to:

[0021] for each type of object attribute information in the at least one type of object attribute information, the following operations are performed respectively:

[0022] for one type of object attribute information, a second weight matrix used for feature extraction on the one type of object attribute information is determined based on a matching association degree between each preset physiological data analysis result and the one type of object attribute information;

[0023] Based on the second weight matrix, the object attribute information of the first type is subjected to feature extraction to obtain a corresponding second intermediate result.

[0024] Optionally, the feature extraction unit is specifically configured to:

[0025] input the original state information of the to-be-analyzed object into a first feature network in the trained target analysis model;

[0026] based on the first feature network, the original state information is subjected to feature extraction to obtain the first intermediate result, wherein a first weight matrix in the first feature network is determined based on the conditional probability.

[0027] Optionally, the feature extraction unit is specifically configured to:

[0028] input at least one type of object attribute information of the to-be-analyzed object into a corresponding second feature network in the target analysis model, wherein each second feature network corresponds to one type of object attribute information;

[0029] based on the respective second feature network corresponding to each type of object attribute information, the corresponding object attribute information is subjected to feature extraction to obtain a corresponding second intermediate result.

[0030] Optionally, the apparatus further comprises:

[0031] a model training unit configured to train the target analysis model by the following manner:

[0032] obtain a sample data set, each sample data in the sample data set comprising a sample object, original state information of the sample object, at least one type of object attribute information, and a physiological data analysis result label of the sample object;

[0033] perform cyclic iteration training on a to-be-trained analysis model according to the sample data set to output a corresponding target analysis model; wherein in one cycle iteration process, the following operations are performed:

[0034] input the selected sample data into the to-be-trained analysis model to obtain a class probability of each type of preset physiological data analysis result corresponding to the sample object;

[0035] adopt a target loss function constructed based on the class probability of each type of preset physiological data analysis result corresponding to the sample object and the physiological data analysis result label to perform parameter adjustment on the to-be-trained analysis model.

[0036] Optionally, in the analysis model to be trained, the first weight matrix in the feature network corresponding to the original state information is initialized based on the conditional probability; and the second weight matrix in the feature network corresponding to each of the at least one type of object attribute information is initialized based on the matching correlation degree between each preset physiological data analysis result and itself.

[0037] The model training unit is specifically configured to:

[0038] Based on the target loss function, the first weight matrix and each second weight matrix are adjusted in parameters.

[0039] Optionally, the model training unit is specifically configured to:

[0040] Divide the initial sample data into at least one sample data subset according to the preset physiological data analysis result categories, and determine the sample quantity in the sample data subset corresponding to each preset physiological data analysis result category;

[0041] Perform up-sampling processing on the sample data in the sample data subset whose sample quantity does not reach the preset quantity, to obtain at least one up-sampled sample data;

[0042] Based on the up-sampled sample data and the initial sample data, construct the sample data set.

[0043] Optionally, the result analysis unit is specifically configured to:

[0044] Determine a summation result obtained by accumulating and summing the first intermediate result and the at least one second intermediate result;

[0045] Perform normalization processing on the summation result, to obtain the category probability of each category of preset physiological data analysis result corresponding to the object to be analyzed.

[0046] Optionally, the object attribute information includes at least one of the following: object age information, object visit time information, and object gender information.

[0047] Optionally, the determining of the target physiological data analysis result of the object to be analyzed based on each category probability includes:

[0048] Taking the preset physiological data analysis result corresponding to the category probability reaching the reference threshold in each category probability as the target physiological data analysis result of the object to be analyzed; or

[0049] Sorting each category probability, and taking the preset physiological data analysis result corresponding to the category probability within the specified order range in the sorting result as the target physiological data analysis result of the object to be analyzed.

[0050] An electronic device provided by an embodiment of the present application includes a processor and a memory, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of any one of the physiological data analysis methods.

[0051] An embodiment of the present application provides a computer readable storage medium, which includes a computer program, and when the computer program runs on an electronic device, the computer program is used to make the electronic device execute the steps of any one of the physiological data analysis methods.

[0052] An embodiment of the present application provides a computer program product, which includes a computer program stored in a computer readable storage medium, and when a processor of an electronic device reads the computer program from the computer readable storage medium, the processor executes the computer program, so that the electronic device executes the steps of any one of the physiological data analysis methods.

[0053] The present application has the following beneficial effects:

[0054] The present application provides a physiological data analysis method, device, electronic device and storage medium. In the present application, the original state information is extracted based on the conditional probability configured according to each preset state information. Each conditional probability represents the probability of occurrence of a preset state information under the condition of a preset physiological data analysis result. Therefore, when the original state information is extracted based on this, the physiological data analysis result more associated with the original state information can be extracted more preferentially, and the accuracy of feature extraction can be improved. In addition, the original state information of the object to be analyzed and the object attribute information related to the physiological data analysis result are combined for multi-class multi-label classification, and the target physiological data analysis result for the object to be analyzed is determined based on the respective category probability of each preset physiological data analysis result. The accuracy is further improved.

[0055] Other features and advantages of the present application will be set forth in the following description, and in part will become apparent to those skilled in the art from the description, or can be learned by practice of the present application. The objects and other advantages of the present application can be realized and achieved by the structure particularly pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF DRAWINGS

[0056] The drawings described herein are used to provide further understanding of the present application, and form a part of the present application. The illustrative embodiments of the present application and their description serve to explain the present application. They do not limit the present application. In the drawings:

[0057] Figure 1 An optional schematic diagram of an application scenario in an embodiment of the present application;

[0058] Figure 2 A schematic diagram of an intelligent inquiry interface in an embodiment of the present application;

[0059] Figure 3 A flowchart of a physiological data analysis method in an embodiment of the present application;

[0060] Figure 4 A schematic diagram of a correspondence between a preset physiological data analysis result and preset state information in an embodiment of the present application;

[0061] Figure 5 A schematic diagram of a processing procedure after inputting original state information and object attribute information of a to-be-analyzed object into a model in an embodiment of the present application;

[0062] Figure 6 A flowchart of a model training method in an embodiment of the present application;

[0063] Figure 7 A structural schematic diagram of an analysis model in an embodiment of the present application;

[0064] Figure 8 A method schematic diagram of sample data expansion in an embodiment of the present application;

[0065] Figure 9A A specific flowchart of a physiological data analysis method in an embodiment of the present application;

[0066] Figure 9B A schematic diagram of a specific scenario of a physiological data analysis method in an embodiment of the present application;

[0067] Figure 10 A component structure schematic diagram of a physiological data analysis apparatus in an embodiment of the present application;

[0068] Figure 11 A hardware component structure schematic diagram of an electronic device applying an embodiment of the present application;

[0069] Figure 12 A hardware component structure schematic diagram of another electronic device applying an embodiment of the present application. DETAILED DESCRIPTION

[0070] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments described in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0071] Some concepts involved in the embodiments of the present application will be introduced below.

[0072] State information: used to represent physiological data of an object in a certain environment. In the embodiments of the present application, the state information is specifically divided into two categories: preset state information and original state information of an object to be analyzed; wherein the original state information represents physiological data of the object to be analyzed in the environment in which the object to be analyzed is located; and the preset state information is historical physiological data of some sample objects determined by statistical analysis or the like.

[0073] Object attribute information: used to represent some characteristic information of the object itself or some characteristic information related to the physiological data analysis result of the object, such as object age, object gender, object visit time, visit history, and the like.

[0074] Physiological data analysis result: refers to a result obtained by analyzing the health status, emotional state and the like of an object based on physiological data of the object. In the embodiments of the present application, the physiological data analysis result can represent emotional fluctuation of the object, such as joy, anger, sadness, and fear; and can also represent physical condition of the object, such as simply represented as healthy, sub-healthy, and unhealthy, or can be specifically subdivided into various diseases. The physiological data analysis result can also be specifically divided into two categories: preset physiological data analysis result and target physiological data analysis result. The preset physiological data analysis result refers to some physiological data analysis results of different categories pre-set in the embodiments of the present application by statistical analysis or the like, such as cold, heat stroke, and the like; and the target physiological data analysis result refers to one or more physiological data analysis results selected from the preset physiological data analysis result.

[0075] Knowledge Graph: also called knowledge domain visualization or knowledge field mapping map, is a series of various different figures showing the development process and structural relationship of knowledge. Knowledge graph uses visualization technology to describe knowledge, and excavates, analyzes, constructs, draws and displays knowledge and their mutual relations. Knowledge graph is a present theory that combines the theories and methods of applied mathematics, graphics, information visualization technology, information science and other disciplines with citation analysis, co-occurrence analysis and other methods, and uses visual map to display the core structure, development history, frontier field and overall knowledge architecture of a discipline to achieve the purpose of multi-disciplinary integration. In the embodiments of the present application, the knowledge graph can be applied in medical diagnosis and intelligent inquiry to provide effective assistance for disease diagnosis.

[0076] Conditional probability: is the probability of event A occurring under the condition that another event B has occurred. Conditional probability is denoted as P(A|B), read as "the probability of A given B". It should be noted that there is no causal or temporal sequence relationship between A and B in these definitions. A may occur before B, or vice versa, or both may occur simultaneously. A may cause the occurrence of B, or vice versa, or there may be no causal relationship between the two. Conditional probability formula, for example, considering the probability of some new information, can be achieved by Bayes' theorem.

[0077] Neural Additive Models (NAM) and feature network: NAM uses neural networks to predict the probability or score of each feature mapping to the final category for each type of feature, and then directly sums these probabilities or scores to obtain the result after regularization. Feature network is a subnetwork in NAM that calculates the probability of each feature for the final category.

[0078] Intermediate result: In the embodiments of the present application, the intermediate result refers to a result obtained before obtaining the final target physiological data analysis result corresponding to the object to be analyzed, which can be divided into two categories: the first intermediate result corresponding to the original state information, and the second intermediate result corresponding to the object attribute information. The first intermediate result specifically represents the probability or score of each preset physiological data analysis result being the analysis result corresponding to the object to be analyzed, which is predicted based on the original state information; the second intermediate result specifically represents the probability or score of each preset physiological data analysis result being the analysis result corresponding to the object to be analyzed, which is predicted based on the object attribute information. Based on the first intermediate result and the second intermediate result, some data processing can be performed to obtain the final target physiological data analysis result corresponding to the object to be analyzed.

[0079] Fine-tune: refers to training the initialized network using data to update the parameters. In the embodiments of the present application, the analysis model needs to be fine-tuned.

[0080] Sample data set: the sample data set in the embodiments of the present application can be divided into two parts: training set and validation set. A machine learning model usually includes two parts of parameters: model parameters and hyperparameters. Among them, the hyperparameters are parameters used to control the behavior of the model, which are not learned by the model itself, for example, the learning rate is a hyperparameter. Among them, the training set is used to adjust the model parameters, and the validation set is used to adjust the hyperparameters, for example, to adjust the learning rate.

[0081] The embodiments of the present application relate to artificial intelligence (AI), natural language processing (NLP) and machine learning technology (ML), and are designed based on computer vision technology and machine learning in artificial intelligence.

[0082] Artificial intelligence is to use digital computers or digital computer controlled machines to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology of computer science, which tries to understand the essence of intelligence and produce a new intelligent machine that can react in a similar way to human intelligence.

[0083] Artificial intelligence is to study the design principles and implementation methods of various intelligent machines, so that the machines have the functions of perception, reasoning and decision-making. Artificial intelligence technology mainly includes computer vision technology, natural language processing technology, and machine learning / deep learning, autonomous driving, intelligent transportation and several other directions. With the research and progress of artificial intelligence technology, artificial intelligence has been researched and applied in many fields, such as common smart home, intelligent customer service, virtual assistant, smart speaker, intelligent marketing, unmanned driving, autonomous driving, robot, intelligent medical treatment, etc. It is believed that with the development of technology, artificial intelligence will be applied in more and more fields and play an increasingly important value.

[0084] Natural language processing is an important direction in the field of computer science and artificial intelligence. It studies various theories and methods that can realize effective communication between people and computers using natural language. Natural language processing is a science that integrates linguistics, computer science and mathematics. Therefore, the research in this field will involve natural language, i.e. the language used in daily life, so it has a close relationship with the study of linguistics. Natural language processing technology usually includes text processing, semantic understanding, machine translation, robot question and answer, knowledge graph and other technologies.

[0085] Machine learning is a multi-disciplinary subject, involving probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory, etc. It is a specialized study of how computers simulate or implement human learning behavior to acquire new knowledge or skills, reorganize existing knowledge structure, and continuously improve their performance. Compared with data mining, which finds mutual characteristics between large data, machine learning focuses more on algorithm design, enabling computers to automatically "learn" rules from data and use the rules to predict unknown data.

[0086] Machine learning is the core of artificial intelligence and the fundamental approach to making computers intelligent. It is applied in various fields of artificial intelligence. Machine learning and deep learning usually include artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, etc.

[0087] The physiological data analysis method proposed in the embodiments of the present application can be divided into two parts, including a training part and an application part; wherein the training part relates to the technical field of machine learning, in the training part, the analysis model is trained by machine learning technology, specifically, the target analysis model is obtained by cyclic iteration training of the analysis model to be trained based on the sample data set. The application part is used to use the target analysis model trained in the training part to combine the original state information and object attribute information of the object to be analyzed to obtain the corresponding target physiological data analysis result.

[0088] The design idea of the embodiments of the present application is briefly introduced as follows:

[0089] Still taking intelligent inquiry as an example, the automatic inquiry system is used to provide online inquiry service to the object. The object can input the symptoms to the automatic inquiry system, and obtain the diseases that may correspond to the symptoms output by the automatic inquiry system, so that the object can make follow-up medical treatment or self-treatment according to the obtained diseases.

[0090] In the related art, the prediction of diseases in the automatic inquiry system is generally realized based on a large amount of labeled data to establish a machine learning model or a neural network model, and the cost required for establishing the machine learning model or the neural network model is high, and the training time is long. The following lists several common machine learning models:

[0091] 1) Bayesian network and its variants. Bayesian network uses a probabilistic graphical model to describe the relationship between variables, and uses a directed graph to describe the joint probability distribution of the variables in the model. Two or three layers of Bayesian network are usually used to model the variables. The specific algorithm is as follows: first, determine the nodes of each layer, and construct the network structure. Statistical data is used to calculate the prior probability of the first layer of nodes, and the conditional probability between each node and all connected parent nodes. When making inferences, the diagnostic inference method is used to calculate the conditional probability of the ancestor nodes from bottom to top. Since the probability distribution of each node combination needs to be known in the Bayesian network, due to the sparsity of data, there may be some node combinations without data, and the probability is 0.

[0092] 2) Naive Bayesian method. When the Bayesian network has only two layers, the first layer is the category and the second layer is the feature, and it is assumed that all features are independent of each other with respect to each category, then it is simplified as the Naive Bayesian method.

[0093] In the method of Bayesian network, the posterior probability is usually used as an indicator for physiological data analysis or sorting, and there is no specific scene for disease diagnosis.

[0094] 3) Machine learning methods such as Support Vector Machines (SVM) and deep neural networks. However, such methods are not interpretable and require a large amount of training data, so they are not widely used in physiological data analysis.

[0095] 4) Knowledge-driven methods. Usually, knowledge cannot be automatically adjusted based on data, and a large amount of manpower is required to manually adjust and correct knowledge, and the results after adjusting each knowledge cannot be viewed in real time, for example, artificially increasing the probability of nasal congestion under the condition of cold, which may lead to some examples of rhinitis being diagnosed as cold, resulting in misdiagnosis.

[0096] In summary, the methods in the related art cannot be applied to the case where only a small amount of data is available. After testing on the validation set, the doctor needs to manually analyze the error causes and then modify them, and after modification, the test is verified again and modified again. This manual iterative process consumes a lot of time for doctors and is very slow.

[0097] In view of this, the embodiment of the present application proposes a physiological data analysis method and device, an electronic device and a storage medium. In the embodiment of the present application, the original state information is extracted based on the conditional probability configured according to each preset state information. Since each conditional probability represents the probability of occurrence of a preset state information under the condition of a preset physiological data analysis result, the physiological data analysis result more associated with the original state information can be extracted when the original state information is extracted based on this, and the accuracy of feature extraction can be improved. In addition, the original state information of the object to be analyzed and the object attribute information related to the physiological data analysis result are combined for multi-class multi-label classification, and the target physiological data analysis result for the object to be analyzed is determined based on the respective category probability of each preset physiological data analysis result. Further improve the accuracy.

[0098] The preferred embodiments of the present application are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present application, and are not used to limit the present application, and the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0099] As shown in the application scenario diagram of the embodiment of the present application. Figure 1 The application scenario diagram includes two terminal devices 110 and a server 120.

[0100] In the embodiment of the present application, the terminal device 110 includes but is not limited to a mobile phone, a tablet computer, a notebook computer, a desktop computer, an electronic book reader, a smart voice interaction device, a smart home appliance, a vehicle-mounted terminal, etc. The terminal device can be installed with a client related to physiological data analysis. The client can be software (such as a browser, smart diagnosis software, health management software, etc.), a webpage, an applet, etc. The server 120 is a background server corresponding to the software or the webpage, the applet, etc., or a server specially used for physiological data analysis. The present application does not make specific limitations. The server 120 can be a standalone physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms, etc. basic cloud computing services.

[0101] It should be noted that the physiological data analysis method in each embodiment of the present application can be executed by an electronic device, which can be the terminal device 110 or the server 120, i.e., the method can be executed by the terminal device 110 or the server 120 alone, or by the terminal device 110 and the server 120 together. For example, when executed by the terminal device 110 and the server 120 together, the terminal device 110 can be used to acquire the original state information input by the subject to be analyzed, and at least one type of object attribute information related to the subject to be analyzed, and send the information to the server 120, and the server 120 can be used to perform feature extraction on the original state information based on the conditional probability configured for each preset state information, to obtain a corresponding first intermediate result; the conditional probability represents the probability of occurrence of a preset state information under the condition of a preset physiological data analysis result; at least one type of object attribute information is respectively extracted to obtain a corresponding second intermediate result of at least one type of object attribute information; finally, the server 120 determines the class probability of each type of preset physiological data analysis result corresponding to the subject to be analyzed based on the first intermediate result and at least one second intermediate result, and determines the target physiological data analysis result of the subject to be analyzed based on each class probability, and then displays the target physiological data analysis result to the subject to be analyzed through the terminal device 110.

[0102] In an optional embodiment, the terminal device 110 and the server 120 can communicate through a communication network.

[0103] In an optional embodiment, the communication network is a wired network or a wireless network.

[0104] It should be noted that, Figure 1 It should be noted that the number of terminal devices and servers is not limited in practice, and is not specifically limited in the embodiments of the present application.

[0105] In the embodiments of the present application, when the number of servers is multiple, the multiple servers can form a block chain, and the servers are nodes on the block chain; the physiological data analysis method disclosed in the embodiments of the present application, wherein the original state information, object attribute information, target physiological data analysis result and other data involved can be saved on the block chain.

[0106] In addition, the physiological data analysis method in the embodiments of the present application can be applied to various physiological data analysis result related scenarios, including but not limited to cloud technology, artificial intelligence (AI), intelligent transportation, assisted driving and the like. In these scenarios, including but not limited to intelligent inquiry tasks, AI assisted diagnosis and treatment tasks, health management detection tasks and the like, can be used for all tasks that require interpretability, have initial weights, and cannot directly train the model with only a small amount of data, but need to adjust the weights according to the data.

[0107] Taking intelligent inquiry as an example, the state information can be represented as a symptom, and the physiological data analysis result can be represented as a disease, as shown in FIG. 2a. Figure 2

[0108] In the embodiments of the present application, the terminal device can provide a disease prediction function for the to-be-analyzed object. When the to-be-analyzed object feels unwell, the disease prediction function in the terminal device can be opened. At this time, the terminal device can guide the to-be-analyzed object to input the related description of the physical discomfort in the object interface, as shown in interface 2a in FIG. 2a. Figure 2 After the terminal device obtains the related description input by the to-be-analyzed object (for example, including the original state information and at least one type of object attribute information), the terminal device can predict the disease information (i.e., the target physiological data analysis result) of the to-be-analyzed object according to the related description, and push the predicted disease information to the to-be-analyzed object, as shown in interface 2b in FIG. 2b. The predicted disease information includes the disease name that the to-be-analyzed object may have. For example, as shown in FIG. 2b, the symptoms of the object are dizziness, fatigue, cough and high body temperature, and the predicted disease is a cold. Figure 2 Figure 2

[0109] It should be noted that, Figure 2 In interface 2a in FIG. 2a, only the related description input by the to-be-analyzed object is taken as an example to illustrate the original state information, Figure 2 the input process of the object attribute information is not shown. In fact, the object attribute information of the to-be-analyzed object can also be input by the object in the current interface, or obtained based on the object account, and the like, which is not limited herein.

[0110] In the embodiments of the present application, the to-be-analyzed object can select a suitable doctor for treatment according to the disease information, avoiding registration errors and repeated registration operations, and can make the to-be-analyzed object timely for treatment, which can greatly save the time cost and the labor cost.

[0111] ​​​It can be understood that in the specific embodiments of the present application, the data related to the original state information of the subject, physiological data, subject attribute information, historical data of the sample subject (such as symptoms, age, gender, time of visit, medical history, etc.), when the above embodiments of the present application are applied to specific products or technologies, the permission or consent of the subject is required, and the collection, use and processing of related data need to comply with relevant laws, regulations and standards of the country and region.

[0112] The physiological data analysis method provided by the example embodiments of the present application will be described below in combination with the application scenarios described above with reference to the accompanying drawings. It should be noted that the above application scenarios are only shown to facilitate understanding of the spirit and principles of the present application, and the embodiments of the present application are not limited in this respect.

[0113] Referring to Figure 3 As shown in FIG. 1, an implementation flowchart of a physiological data analysis method provided by an embodiment of the present application, the specific implementation flow of the method includes the following steps S31-S34:

[0114] S31: The server obtains the original state information of the subject to be analyzed, and at least one type of subject attribute information related to the subject to be analyzed.

[0115] The original state information includes the physiological data of the subject to be analyzed in the environment in which the subject to be analyzed is located, such as symptom information. The symptom information can represent a series of abnormal changes in function, metabolism and morphological structure in the body during the disease process, which causes abnormal subjective feelings or certain objective pathological changes of the patient.

[0116] In the embodiments of the present application, the symptoms include two aspects of symptoms and signs, which refer to subjective discomfort, abnormal feeling, functional change or obvious pathological change of the patient caused by the disease. Some common symptoms are: fever, pain, weight change, edema, dyspnea, cough, hemoptysis, loss of appetite, indigestion, dysphagia, nausea and vomiting, hematemesis, anemia, shock, etc. In addition, it should be noted that the symptoms can also include some information determined in combination with the subjective perception of the patient, such as the number of vomiting, the color of vomiting, etc., which are all some physiological data of the subject in the environment in which the subject is located.

[0117] It should be noted that the above only simply lists some symptoms, and in fact any state information that can represent the physiological data of the subject in a certain environment is applicable to the embodiments of the present application, which is not specifically limited herein.

[0118] In addition, the subject attribute information represents at least one type of feature information related to the acquisition of physiological data analysis results, such as subject age information, subject gender information, subject visit time information, etc. In addition, it can further include medical history information, etc., which is not specifically limited herein.

[0119] S32: The server performs feature extraction on the original state information based on the conditional probability configured corresponding to each preset state information, to obtain a corresponding first intermediate result.

[0120] The conditional probability represents the probability of occurrence of a preset state information under the condition of a preset physiological data analysis result. Since each preset state information can represent some different state information, for example, different symptoms, and each preset physiological data analysis result refers to some different categories of results, for example, different diseases, the conditional probability can also represent the probability of occurrence of a symptom under the condition that a disease has occurred, for example, the probability of nausea under the condition of heat stroke; the probability of dizziness under the condition of heat stroke; the probability of cough under the condition of a cold; and so on.

[0121] In the embodiments of the present application, the conditional probability configured corresponding to each preset state information can be obtained according to books, doctor's experience, knowledge graph or other channels, for example, there are many well-organized knowledge graphs in the medical and health care field, such as international disease classification, clinical guidelines and consensus, etc., which have hierarchical information and complex correlation consistent with human cognition, and the present application is not limited in this regard.

[0122] Referring to Figure 4 , which is a schematic diagram of the correspondence between a preset physiological data analysis result and a preset state information in the embodiments of the present application.

[0123] In the embodiments of the present application, the conditional probability configured corresponding to each preset state information can be obtained according to books, doctor's experience, knowledge graph or other channels, for example, there are many well-organized knowledge graphs in the medical and health care field, such as international disease classification, clinical guidelines and consensus, etc., which have hierarchical information and complex correlation consistent with human cognition, and the present application is not limited in this regard. Figure 4 Some common preset state information (symptoms) corresponding to preset physiological data analysis results (diseases) are simply listed. For example, common symptoms of a cold include nasal congestion, runny nose, throat itching, sore throat, cough, headache, dizziness, aversion to cold, low fever, etc.; common symptoms of hypoglycemia include palpitations, hand tremors, sweating, headaches, dizziness, coma, convulsions, etc.; common symptoms of acute gastroenteritis include nausea, vomiting, abdominal pain, diarrhea, fever, thirst, etc.; common symptoms of heat stroke include profuse sweating, dizziness, blurred vision, nausea, palpitations, shortness of breath, and body temperature often less than 37.5 degrees Celsius, etc.

[0124] It should be noted that the above-mentioned correspondence is only a simple example, and the correspondence between any preset physiological data analysis result and preset state information is applicable to the embodiments of the present application, and is not limited in this regard.

[0125] In addition, in the embodiments of the present application, the manner of obtaining the correspondence between the preset physiological data analysis result and the preset state information is the same as or similar to the manner of obtaining the conditional probability described above, and the repeated parts will not be described again.

[0126] An optional implementation is that step S32 can be further divided into the following sub-stepsFigure 3 not shown):

[0127] S321: Obtain the conditional probability of each preset state information under each preset physiological data analysis result respectively, and determine a first weight matrix for feature extraction based on the obtained conditional probability.

[0128] Specifically, a conditional probability matrix composed of the conditional probability of each symptom under each disease is obtained from the knowledge graph, a first weight matrix is initialized based on the conditional probability matrix, and the first weight matrix is updated through machine learning, and finally a first weight matrix for feature extraction is obtained.

[0129] For example, given a disease set D = {d1, d2,... d N}, where N is the number of diseases (i.e., the number of categories), d i is a certain disease, such as common cold, and i takes values from 1 to N. The symptom feature set F = {f1, f2,... f K}, where K is the number of symptom features, f j is a certain feature, such as cough, and each feature takes values from {-1, 0, 1}, representing negative symptoms, no mention, and positive symptoms, respectively, and j takes values from 1 to K.

[0130] According to books, doctor experience or other channels, the conditional probability matrix C of each symptom feature under each disease is obtained, which has a size of K x N, where each element C ji takes a value of P(f j |d i ) and a value range of [0, 1].

[0131] Where, when a first weight matrix is initialized based on the conditional probability matrix, the initialization method here can be many, such as directly assigning values to the elements in the first weight matrix based on the conditional probability matrix, or using mapping, etc., which is not limited in this paper, and the size of the first weight matrix is also K x N.

[0132] S322: Based on the first weight matrix, the original state information is subjected to feature extraction to obtain a corresponding first intermediate result.

[0133] Specifically, the original state information of the object to be analyzed can be input into the first feature network of the trained target analysis model; the original state information is subjected to feature extraction based on the first feature network to obtain a first intermediate result, wherein the first weight matrix in the first feature network is determined based on the conditional probability, and the process is the same as the model training process, and the specific calculation method can be referred to formula (1-1) below, and the repeated parts will not be described again.

[0134] In the above embodiment, the knowledge is represented as the conditional probability of each symptom, and the experts are invited to label the probability according to books and experience, and the feature network is initialized using the labeled probability, so that the accuracy can be effectively improved.

[0135] S33: The server respectively performs feature extraction on the at least one type of object attribute information, and obtains a second intermediate result corresponding to each of the at least one type of object attribute information.

[0136] An optional embodiment is that step S33 can be further divided into the following substeps (not shown): Figure 3

[0137] For each of the at least one type of object attribute information, the following operations are respectively performed:

[0138] S331: For a type of object attribute information, a second weight matrix used for feature extraction of the type of object attribute information is determined based on a matching association degree between each preset physiological data analysis result and the type of object attribute information.

[0139] Similar to the first weight matrix, a weight matrix can be initialized based on the matching association degree between each preset physiological data analysis result and the type of object attribute information, and the weight matrix is updated through machine learning, and finally a second weight matrix used for feature extraction is obtained.

[0140] Taking that the object attribute information includes object age information as an example, the matching association degree between each type of disease and the object age information indicates the association between each type of disease and age, for example, for hand-foot-mouth disease, the main crowd of onset is children under the age of 5; for the elderly (over 60 years old), the high-incidence diseases include coronary heart disease; for middle-aged people (45-59 years old), the high-incidence diseases include hypertension. That is, the matching association degree between hand-foot-mouth disease and the age range of 0-5 years old is higher, the matching association degree between coronary heart disease and the age range of over 60 years old is higher, and the matching association degree between hypertension and the age range of 45-59 years old is higher.

[0141] For another example, taking that the object attribute information includes object treatment time information as an example, the matching association degree between each type of disease and the object treatment time information indicates the association between each type of disease and time, for example, heatstroke frequently occurs in summer; influenza frequently occurs in spring. That is, the matching association degree between heatstroke and the summer time period is higher, and the matching association degree between influenza and the spring time period is higher.

[0142] When the second weight matrix is initialized based on the matching association degree, a higher weight can be set when the matching association degree is higher, which can be set as a positive value, for example, 2, and a lower weight can be set when the matching association degree is lower, which can be set as a negative value, for example, -2.​

[0143] It should be noted that the above-described methods for determining the second weight matrix based on the degree of matching association are merely illustrative examples. In fact, any method for determining the second weight matrix based on the degree of matching association is applicable to the embodiments of this application, and no specific limitation is made here.

[0144] In addition, the second weight matrix can be randomly initialized. Compared with random initialization, the method of determining the weight matrix based on the degree of matching association can more closely associate the preset physiological data analysis results with the object attribute information. Therefore, the accuracy of the analysis results obtained in this way is higher.

[0145] S332: Based on the second weight matrix, feature extraction is performed on the attribute information of a class of objects to obtain the corresponding second intermediate result.

[0146] Specifically, at least one type of object attribute information of the object to be analyzed can be input into the corresponding second feature network in the target analysis model, wherein each second feature network corresponds to one type of object attribute information; then, based on the corresponding second feature network of at least one type of object attribute information, feature extraction is performed on the corresponding object attribute information to obtain the corresponding second intermediate result. This process is the same as the model training process. For the specific calculation method, please refer to the following formula (1-2) or (1-3). Repeated parts will not be repeated.

[0147] like Figure 5 The diagram illustrates a processing procedure in this application embodiment where the original state information and object attribute information of the object to be analyzed are input into the target analysis model. Taking object attribute information including object age information and object visit time information as an example, the second intermediate result includes: the second intermediate result corresponding to the object age information and the second intermediate result corresponding to the object visit time information.

[0148] like Figure 5 In this process, the original state information of the object to be analyzed is input into the first feature network to obtain the first intermediate result; the object's age information is input into the corresponding second feature network-1 to obtain the second intermediate result-1; and the object's medical visit time information is input into the corresponding second feature network-2 to obtain the second intermediate result-2.

[0149] Next, we can proceed to step S34:

[0150] S34: Based on the first intermediate result and at least one second intermediate result, the server determines the category probability of the pre-defined physiological data analysis results corresponding to the object to be analyzed, and determines the target physiological data analysis result of the object to be analyzed based on the category probability.

[0151] Optionally, when the object attribute information includes the object age information and the object visit time information, the class probability can be determined by the following manner:

[0152] First, a summation result obtained by accumulating and summing the first intermediate result and the at least one second intermediate result is determined; then, the summation result is normalized to obtain the class probability of each preset physiological data analysis result corresponding to the object to be analyzed.

[0153] As shown in the following formula (1), the first intermediate result, the second intermediate result-1 and the second intermediate result-2 are accumulated, and then the summation result is normalized by the activation function Sigmoid function to obtain the class probability of each preset physiological data analysis result corresponding to the object to be analyzed. Figure 5

[0154] For example, taking the total number N of preset physiological data analysis results as 4, i.e., four types of preset physiological data analysis results are set, which are cold, hypoglycemia, acute gastroenteritis and heatstroke; the original state information of the object to be analyzed includes dizziness, fatigue, cough and high body temperature; the age is 20 years old; based on this, the class probabilities corresponding to cold, hypoglycemia, acute gastroenteritis and heatstroke of the object to be analyzed are respectively P1, P2, P3 and P4.

[0155] For another example, when the object attribute information further includes the object gender information, the second intermediate result further includes the second intermediate result corresponding to the object gender information. At this time, the summation result refers to the result obtained by accumulating and summing the first intermediate result, the second intermediate result corresponding to the object age information, the second intermediate result corresponding to the object visit time information and the second intermediate result corresponding to the object gender information, and so on. When the object attribute information is changed, the same method can be used, which will not be repeated here.

[0156] Based on the above manner, the class probability corresponding to each preset physiological data analysis result can be obtained, and based on each class probability, one or more target physiological data analysis results matched with the object to be analyzed can be determined.

[0157] An optional embodiment is that there are two ways to determine the target physiological data analysis result of the object to be analyzed based on each class probability:

[0158] The first determination method is to determine the preset physiological data analysis result corresponding to the class probability reaching the reference threshold value in each class probability as the target physiological data analysis result of the object to be analyzed.

[0159] ​For example, when the reference threshold is tr, if the category probability of a certain disease is greater than the threshold, the result of the matched target physiological data analysis is obtained; assuming that the four category probabilities P1, P2, P3, and P4 listed above are P1>tr, it indicates that the disease that the object to be analyzed may have is a cold.

[0160] In a second determination manner, the category probabilities are sorted, and the preset physiological data analysis result corresponding to the category probability in the specified order range is taken as the target physiological data analysis result of the object to be analyzed.

[0161] For example, when the category probabilities are sorted from large to small, the first m can be selected as the target physiological data analysis result of the object to be analyzed; when the category probabilities are sorted from small to large, the last m can be selected as the target physiological data analysis result of the object to be analyzed, and m is a positive integer.

[0162] For example, P1>P3>P2>P4, and m=2, which indicates that the disease that the object to be analyzed may have is at least one of a cold and acute gastroenteritis, and the most likely disease is a cold.

[0163] It should be noted that the above-mentioned several determination manners are only examples, and any manner of determining the target physiological data analysis result of the object to be analyzed based on the category probabilities is applicable to the embodiments of the present application, and repeated descriptions are omitted.

[0164] In the embodiments of the present application, the feature extraction process can be realized by machine learning, for example, the analysis model for physiological data analysis result matching proposed in the present application.

[0165] Considering that the medical diagnosis scene needs high interpretability and needs to be perfectly compatible with the meaning of conditional probability, the NAM model is used for modeling. Taking the analysis model as an example, for each feature (i.e., original state information, object attribute information, etc.), a feature network can be used to calculate the network output h of the feature, and then, based on the output of each feature network, N category probabilities are predicted.

[0166] In the embodiments of the present application, the sample data set used for training the NAM can be divided into two parts, a training set and a validation set. The training set is used to train the NAM, and for each iteration, the loss function Loss is calculated, and the gradient of Loss is calculated. The optimizer is used to update the parameters. After the iteration training is completed, the model is saved. Then, the samples of the validation set are input, the Loss is calculated, and the required indicators are calculated. According to a certain training strategy, the learning rate is adjusted, and it is determined whether to terminate early.

[0167] For example, the preset number of iterations is 20, but after the 5th iteration, the Loss calculated based on the validation set no longer decreases, but is in an oscillating state, and the iteration can be stopped in advance.

[0168] After the training is completed, the probabilities P of each category are calculated by the optimal model as the output of the model. Then the diseases are sorted according to P or the diseases with a probability Pi higher than a preset threshold are output as the result.

[0169] The detailed process of model training is illustrated as follows:

[0170] An optional implementation is that the target analysis model can be obtained by training in the following way:

[0171] Referring to Figure 6 , which is a training process diagram of a target analysis model in the embodiment of the application, taken the server as an execution subject, specifically includes the following steps S61-S62:

[0172] S61: The server acquires a sample data set, each sample data in the sample data set including a sample object, original state information of the sample object, at least one type of object attribute information, and a physiological data analysis result label of the sample object.

[0173] For example, for each sample data S l (l is valued at 1-n), a known feature sequence F l (that is, the original state information) and a disease label D l (that is, the physiological data analysis result label) are known, F l is a subset of F, wherein each element is a symptom feature observed in the sample object, in addition to object attribute information such as age, gender, and visit month. Here, the symptom feature is taken as an example for initialization and fine-tuning. D l is a subset of the disease set D, representing the true disease. Since the number of true diseases is not limited in the embodiment of the application, it is a multi-label problem of predicting disease probability given the known feature sequence.

[0174] Specifically, for each sample data, the feature sequence F l can be mapped into a vector of length K, wherein the value of a positive feature is 1, the value of a negative feature is -1, and the value of a feature not observed is 0. The disease label is mapped into a vector of length N, wherein the value of a true disease is 1 and the values of the rest are 0. Thus, a vectorized training set representation is obtained. The same mapping method is used for the validation set, and each sample is represented as a vector pair to obtain a vectorized validation set representation.

[0175] On the basis of the sample data containing the above-mentioned content, a small data amount training set S train and a validation set Svalid Each set S = {(F1,D1),(F2,D2),...,(F...} n D n The goal of training this model is to adjust the conditional probability matrix C to improve the model's prediction accuracy.

[0176] S62: The server iteratively trains the analysis model to be trained based on the sample dataset and outputs the corresponding target analysis model.

[0177] In one iteration, the following operations are performed:

[0178] S621: The server inputs the selected sample data into the analysis model to be trained to obtain the category probability of the sample object corresponding to various preset physiological data analysis results.

[0179] Specifically, this process is the same as the process described above for analyzing physiological data of the object to be analyzed based on the target analysis model. For example... Figure 7 As shown, this is a schematic diagram of an analysis model structure in an embodiment of this application. This analysis model is an additive neural network, where each type of feature corresponds to a feature network. The feature network can have one or more layers, and the number of neurons in the last layer is the number of diseases, N. For example... Figure 7 In the feature network 1, ..., feature network n, the various features input into the analysis model are fed into their respective feature networks for feature extraction. Figure 7 The feature vector formed by the original state information, object attribute information, etc. is input into the corresponding feature network 1-feature network n in segments. Finally, the hidden layer outputs of each feature network are accumulated and summed, and then normalized by Sigmoid to obtain the probability vector P composed of the probabilities of each category, which contains N elements. Each element represents the category probability corresponding to a preset physiological data analysis result (disease).

[0180] In this embodiment, the features (i.e., original state information, object attribute information, etc.) and labels (i.e., physiological data analysis result labels) of the training and validation sets are vectorized and input into an additive neural network, such as... Figure 7 As shown.

[0181] The following example illustrates this, using the original state information as symptoms and the object attribute information as age and consultation time:

[0182] For symptom characteristics x s Using a one-layer neural network, where the first weight matrix W s Initialize using the labeled conditional probability matrix C.

[0183] During training and inference, the output h of the feature network is calculated according to formula (1-1). s The first intermediate result is a vector of length N.

[0184] h s =g(W s x s +b s ) Formula (1-1)

[0185] Among them, W s b is the first trainable weight matrix, with the same size as C. s is the bias vector with length N. g is the activation function used to regularize the output. Commonly used activation functions include tanh, ReLU, Leaky ReLU, etc., and different activation functions can be selected according to different tasks.

[0186] Specifically, this application takes into account that if a rectified linear unit (ReLU) or sigmoid is used, the scores of each category after the subsequent sigmoid transformation will be above 0.5. Therefore, in this disease probability prediction task, tanh or a leaky rectified linear unit (Leaky ReLU) is used.

[0187] Similarly, for age characteristic x a This refers to the second intermediate result corresponding to the object's age information, which is also a vector of length N, and can be calculated according to formula (1-2):

[0188] h a =g(W a x a +b a ) Formula (1-2)

[0189] Among them, h a W is the hidden output of the age feature network. a b is the trainable second weight matrix. a This is the bias vector.

[0190] Regarding the characteristic of consultation time x m This refers to the second intermediate result corresponding to the patient's consultation time information, which is also a vector of length N, and can be calculated according to formula (1-3):

[0191] h m =g(W m x m +b m ) Formula (1-3)

[0192] Among them, h mW m is also a trainable second weight matrix, and b m is a bias vector.

[0193] Subsequently, the vectors output by each feature network can be added and normalized using a Sigmoid function to obtain the probability P of each class, as shown in equation (2):

[0194] P = f(∑(h s ,h a ,h m )) equation (2)

[0195] where P is a vector of length N, each element P i is the probability of belonging to the corresponding class, i.e., the class probability. f is a normalization function that maps the value range to [0, 1]. Considering that the embodiments of the present application are multi-label classification, a Sigmoid function can be used as f (if it is a single classification problem, functions such as softmax can be used as f).

[0196] S622: The server uses the class probability and the physiological data analysis result label based on the sample object corresponding to each class of preset physiological data analysis result to construct a target loss function for the analysis model to be trained to adjust the parameters.

[0197] For the training set, the network output of all samples and the label vector are calculated to calculate the loss function. Common loss functions for classification problems include cross-entropy and its variants. Taking cross-entropy as an example, the calculation of the loss function is shown in equation (3).

[0198] Loss = -∑ i y i logP i -(1-y i )log(1-P i ) equation (3)

[0199] In equation (3), y i is the actual probability of each class (i.e., the physiological data analysis result label in the sample data), which is 0 or 1, and P i is the predicted probability of the class, which is the i-th element of the probability matrix P.

[0200] Specifically, the gradient of Loss is calculated during training, and any optimization method is used for parameter update. After each iteration, the error and accuracy on the validation set data are calculated, and the model is saved until the preset number of iterations is reached to end the training. The training can also be ended early as needed, such as when the error no longer decreases or the accuracy no longer increases, the model with the highest accuracy or the smallest error is taken as the disease prediction model, i.e., the target analysis model.

[0201] It should be noted that in the above embodiment, an interpretable high-accuracy method, an additive neural network, is used to predict disease probability, and the initialized additive neural network is fine-tuned using historical data of some sample objects to improve accuracy. This process automatically adjusts the parameters of the neural network, reduces manual labor, effectively improves efficiency, and achieves the effect of using NAM for multi-class multi-label classification.

[0202] Optionally, in the embodiments of the present application, after training, the influence of each feature on each category and the parameters can be visualized or printed to check whether the relationship and parameter values are reasonable, so as to adjust the network or parameters, and some features can also be removed, which is specific to actual conditions and is not limited here.

[0203] In the embodiments of the present application, if there is an imbalance between categories in the training set or the validation set, the class with less data can be up-sampled to make the sample size of each category the same.

[0204] An optional implementation is to expand the sample data in the following way:

[0205] First, the initial sample data is divided into at least one sample data subset according to the preset physiological data analysis result category, and the number of samples in the sample data subset corresponding to each preset physiological data analysis result category is determined; then, the sample data in the sample data subset whose sample number does not reach the preset number is up-sampled to obtain at least one up-sampled sample data. In this way, based on the up-sampled sample data and the initial sample data, a sample data set can be constructed.

[0206] As shown in Figure 8 , it is a method schematic diagram of sample data expansion in the embodiments of the present application, wherein the sample data set contains initial sample data and up-sampled sample data, wherein the initial sample data contains four categories of sample data subsets divided according to the preset physiological data analysis result category, which are represented by rectangles filled with different patterns in Figure 8 . Among them, the sample data in the last two categories of sample data subsets is less, which does not reach the preset number, so up-sampling can be performed to obtain the up-sampled sample data in the dashed box. Finally, after the sample data is expanded in the above way, it can be input into the analysis model for model training and verification, and finally the target analysis model is output.

[0207] Among them, when up-sampling, the sample data can be randomly sampled several times according to certain rules to increase the data size. For example, the age in the object attribute information, the adjustment of the treatment time, etc. are not limited here.

[0208] After expanding the sample data in the above manner, a large amount of sample data (labeled sample data) can be obtained. Then, sample data can be extracted from the sample dataset. For example, the extracted sample data includes the original state information of the sample object, the age information of the object, the time information of the object's medical treatment, and the labels of the physiological data analysis results. Then, the model can be trained and validated to obtain the target analysis model.

[0209] In the above embodiments, for scenarios with only a small amount of data, the above method can also be used to expand the sample and then train the model, which can solve the problem in related technologies that the model cannot be directly trained with only a small amount of data.

[0210] In summary, the embodiments of this application initialize the feature network using the conditional probabilities of each symptom and fine-tune it using a training set with a very small amount of data. This allows for automatic adjustment of the conditional probabilities, improving accuracy and reducing manual labor.

[0211] See Figure 9A The diagram shown is a schematic flowchart of a physiological data analysis method according to an embodiment of this application. Taking the server as the execution entity as an example, the specific implementation flow of this method is as follows:

[0212] Step S901: The server obtains the original state information of the object to be analyzed, as well as the object's age information and the object's medical visit time information;

[0213] Step S902: The server inputs the original state information into the first feature network of the target analysis model, and inputs the object's age information and the object's medical visit time information into the corresponding second feature network of the target analysis model, respectively.

[0214] Step S903: The server extracts features from the original state information based on the first feature network to obtain the first intermediate result;

[0215] Step S904: The server extracts features from the corresponding object attribute information based on the corresponding second feature network to obtain the second intermediate result corresponding to the object's age information and the second intermediate result corresponding to the object's medical visit time information.

[0216] Step S905: Based on the first intermediate result, the second intermediate result corresponding to the object's age information, and the second intermediate result corresponding to the object's medical visit time information, the server determines the summation result obtained by accumulating and summing.

[0217] Step S906: The server normalizes the summation result to obtain the category probability of the analysis results of various preset physiological data corresponding to the object to be analyzed;

[0218] Step S907: The server uses the preset physiological data analysis results corresponding to the category probabilities that reach the reference threshold in each category as the target physiological data analysis results for the object to be analyzed.

[0219] like Figure 9B The diagram illustrates a specific scenario of a physiological data analysis method described in this application embodiment. The original state information of the subject to be analyzed includes: sweating, dizziness, blurred vision, and nausea; the subject's age is 20 years old; and the subject's medical visit date is July 2nd. After the subject inputs this information through a client installed on a terminal device, the terminal device can send this information to a server. The server is configured with a target analysis model. Based on the feature networks in the target analysis model, feature extraction is performed to obtain intermediate results, including a first intermediate result corresponding to the original state information, a second intermediate result corresponding to the subject's age information, and a second intermediate result corresponding to the subject's medical visit date. Then, using the methods described in steps S905-S907 above, the target physiological data analysis results of the subject to be analyzed are obtained and fed back to the terminal device, which then displays the results to the subject through the client. For example, the server, through the target analysis model, ultimately obtains the probabilities of each category corresponding to the object to be analyzed as follows: heatstroke probability 0.7, acute gastroenteritis probability 0.5, cold probability 0.4, hypoglycemia probability 0.3. The reference threshold is 0.6, so the probability of heatstroke reaching the reference threshold is 0.7. The corresponding preset physiological data analysis result is: heatstroke. This result can then be fed back to the terminal device, which can then display the result to the object to be analyzed, such as... Figure 9B The intelligent consultation interface shown.

[0220] Combining the physiological data analysis method in the embodiments of this application, in the pediatric intelligent consultation scenario, a total of 110 diseases were tested. Using a dataset of 719 samples, the parameters were fine-tuned, and the accuracy on the validation set increased from 26.4% to 51.7%, an improvement of 95.8%, with an even greater improvement on P@3.

[0221] Based on the same inventive concept, embodiments of this application also provide a device for analyzing physiological data. For example... Figure 10 As shown, this is a schematic diagram of the structure of a physiological data analysis device 1000, which may include:

[0222] The information acquisition unit 1001 is used to acquire the original state information of the object to be analyzed, as well as at least one type of object attribute information related to the object to be analyzed; wherein, the original state information includes the physiological data of the object to be analyzed in its own environment.

[0223] The feature extraction unit 1002 is configured to perform feature extraction on the original state information based on conditional probabilities corresponding to the preset state information configurations, to obtain corresponding first intermediate results; the conditional probability represents a probability of occurrence of a preset state information under a preset physiological data analysis result.

[0224] The feature extraction unit 1002 is further configured to perform feature extraction on at least one type of object attribute information respectively, to obtain corresponding second intermediate results of the at least one type of object attribute information respectively.

[0225] The result analysis unit 1003 is configured to determine a class probability of each type of preset physiological data analysis result of the to-be-analyzed object based on the first intermediate result and the at least one second intermediate result, and determine a target physiological data analysis result of the to-be-analyzed object based on the class probabilities.

[0226] Optionally, the feature extraction unit 1002 is specifically configured to:

[0227] The conditional probabilities of the preset state information under each preset physiological data analysis result are obtained respectively, and a first weight matrix used for feature extraction is determined based on the obtained conditional probabilities.

[0228] The original state information is subjected to feature extraction based on the first weight matrix, to obtain the corresponding first intermediate result.

[0229] Optionally, the feature extraction unit 1002 is specifically configured to:

[0230] For each type of object attribute information in the at least one type of object attribute information, the following operations are performed respectively:

[0231] For a type of object attribute information, a second weight matrix used for feature extraction on the type of object attribute information is determined based on a matching association degree between each preset physiological data analysis result and the type of object attribute information.

[0232] The type of object attribute information is subjected to feature extraction based on the second weight matrix, to obtain the corresponding second intermediate result.

[0233] Optionally, the feature extraction unit 1002 is specifically configured to:

[0234] The original state information of the to-be-analyzed object is input into a first feature network of a trained target analysis model.

[0235] The original state information is subjected to feature extraction based on the first feature network, to obtain the first intermediate result, wherein a first weight matrix in the first feature network is determined based on the conditional probability.

[0236] Optionally, the feature extraction unit 1002 is specifically configured to:

[0237] input at least one type of object attribute information of the object to be analyzed into a corresponding second feature network in the target analysis model, wherein each second feature network corresponds to one type of object attribute information;

[0238] perform feature extraction on the corresponding object attribute information based on the respective second feature network of each type of object attribute information to obtain a corresponding second intermediate result.

[0239] Optionally, the device further comprises:

[0240] a model training unit 1004 configured to train the target analysis model by the following manner:

[0241] obtain a sample data set, each sample data in the sample data set comprising a sample object, original state information of the sample object, at least one type of object attribute information, and a physiological data analysis result label of the sample object;

[0242] perform cyclic iteration training on the analysis model to be trained based on the sample data set to output a corresponding target analysis model; wherein in one cycle iteration process, the following operations are performed:

[0243] input the selected sample data into the analysis model to be trained to obtain a class probability of each type of preset physiological data analysis result corresponding to the sample object;

[0244] adjust the parameters of the analysis model to be trained using a target loss function constructed based on the class probability of each type of preset physiological data analysis result corresponding to the sample object and the physiological data analysis result label.

[0245] Optionally, in the analysis model to be trained, the first weight matrix in the feature network corresponding to the original state information is initialized based on a conditional probability; and the second weight matrix in the feature network corresponding to each type of object attribute information is initialized based on a matching correlation degree between each preset physiological data analysis result and itself.

[0246] The model training unit 1004 comprises the following units:

[0247] adjust the parameters of the first weight matrix and each second weight matrix based on the target loss function.

[0248] Optionally, the model training unit 1004 is specifically configured to:

[0249] divide the initial sample data into at least one sample data subset according to the preset physiological data analysis result categories, and determine the sample quantity in the sample data subset corresponding to each preset physiological data analysis result category;

[0250] The sample data in the sample data subset that does not reach the preset number of samples is up-sampled to obtain at least one up-sampled sample data;

[0251] Based on the up-sampled sample data and the initial sample data, the sample data set is constructed.

[0252] Optionally, the result analysis unit 1003 is specifically configured to:

[0253] determine a summation result obtained by accumulating and summing the first intermediate result and the at least one second intermediate result;

[0254] perform normalization processing on the summation result to obtain a class probability of each type of preset physiological data analysis result corresponding to the to-be-analyzed object.

[0255] Optionally, the object attribute information includes at least one of the following: object age information, object visit time information, and object gender information.

[0256] Optionally, based on the class probabilities, a target physiological data analysis result of the to-be-analyzed object is determined, including:

[0257] a preset physiological data analysis result corresponding to a class probability that reaches a reference threshold in the class probabilities is taken as the target physiological data analysis result of the to-be-analyzed object; or

[0258] the class probabilities are sorted, and a preset physiological data analysis result corresponding to a class probability in a specified order range of the sorting result is taken as the target physiological data analysis result of the to-be-analyzed object.

[0259] In the embodiments of the present application, the original state information is feature-extracted based on the conditional probabilities configured corresponding to each preset state information. Since each conditional probability represents the probability of occurrence of a preset state information under the condition of a preset physiological data analysis result, when the original state information is feature-extracted based on this, the physiological data analysis result more associated with the original state information can be more preferentially extracted, and the accuracy of feature extraction can be improved. In addition, the original state information of the to-be-analyzed object and the object attribute information related to the physiological data analysis result are combined for multi-class multi-label classification, and the target physiological data analysis result for the to-be-analyzed object is determined based on the class probabilities corresponding to each preset physiological data analysis result. The accuracy is further improved.

[0260] For the convenience of description, each part is described as a module (or unit) according to function. Of course, the functions of the modules (or units) can be implemented in the same or multiple software or hardware when the present application is implemented.

[0261] After introducing the physiological data analysis method and device of the exemplary embodiments of the present application, next, the electronic device according to another exemplary embodiment of the present application is introduced.

[0262] Those skilled in the art can understand that various aspects of the present application can be implemented as a system, a method or a program product. Therefore, various aspects of the present application can be embodied as a complete hardware embodiment, a complete software embodiment (including firmware, microcode, etc.), or an embodiment combining hardware and software aspects, which can be collectively referred to as "circuitry", "module" or "system" herein.

[0263] Based on the same inventive concept as the above method embodiments, the present embodiments also provide an electronic device. In an embodiment, the electronic device can be a server, such as the server 120 shown in Figure 1 In this embodiment, the structure of the electronic device can be as shown in Figure 11 The electronic device includes a memory 1101, a communication module 1103 and one or more processors 1102.

[0264] The memory 1101 is used to store computer programs executed by the processor 1102. The memory 1101 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system and programs required for running instant messaging functions, etc.; the data storage area can store various instant messaging information and operation instruction sets, etc.

[0265] The memory 1101 can be a volatile memory such as a random-access memory (RAM); the memory 1101 can also be a non-volatile memory such as a read-only memory, a flash memory, a hard disk drive (HDD) or a solid-state drive (SSD); or the memory 1101 can be any other medium capable of carrying or storing desired computer programs in the form of instructions or data structures and capable of being accessed by a computer, but is not limited to this. The memory 1101 can be a combination of the above memories.

[0266] The processor 1102 can include one or more central processing units (CPUs) or digital processing units, etc. The processor 1102 is used to implement the above physiological data analysis method when invoking the computer programs stored in the memory 1101.

[0267] The communication module 1103 is used to communicate with terminal devices and other servers.

[0268] This application embodiment does not limit the specific connection medium between the memory 1101, communication module 1103, and processor 1102. This application embodiment... Figure 11 The memory 1101 and the processor 1102 are connected via a bus 1104, and the bus 1104 is in Figure 11 The diagram uses thick lines to describe the connections between other components; these are for illustrative purposes only and should not be considered limiting. Bus 1104 can be divided into address bus, data bus, control bus, etc. For ease of description, Figure 11 It is described using only a thick line, but does not indicate that there is only one bus or one type of bus.

[0269] The memory 1101 stores a computer storage medium, which stores computer-executable instructions for implementing the physiological data analysis method of this application embodiment. The processor 1102 is used to execute the aforementioned physiological data analysis method, such as... Figure 3 As shown.

[0270] In another embodiment, the electronic device may also be other electronic devices, such as... Figure 1 The terminal device 110 is shown. In this embodiment, the electronic device can be structured as follows: Figure 12 As shown, it includes components such as: communication component 1210, memory 1220, display unit 1230, camera 1240, sensor 1250, audio circuit 1260, Bluetooth module 1270, processor 1280, etc.

[0271] The communication component 1210 is used to communicate with the server. In some embodiments, it may include a Wireless Fidelity (WiFi) module, which is a short-range wireless transmission technology. Electronic devices can use the WiFi module to help objects send and receive information.

[0272] The memory 1220 can be used to store software programs and data. The processor 1280 executes various functions of the terminal device 110 and performs data processing by running the software programs or data stored in the memory 1220. The memory 1220 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device. The memory 1220 stores an operating system that enables the terminal device 110 to run. In this application, the memory 1220 can store the operating system and various application programs, and may also store computer programs that execute the physiological data analysis methods of the embodiments of this application.

[0273] The display unit 1230 can also be used to display information input by the object or information provided to the object, as well as the graphical object interface of various menus of the terminal device 110. Specifically, the display unit 1230 can include a display screen 1232 arranged on the front of the terminal device 110. Among them, the display screen 1232 can be configured in the form of a liquid crystal display, a light-emitting diode, etc. The display unit 1230 can be used to display the application operation interface (such as the intelligent inquiry interface shown in the embodiments of the present application) in the present application. Figure 2

[0274] The display unit 1230 can also be used to receive input digital or character information, generate signal input related to object settings and function control of the terminal device 110. Specifically, the display unit 1230 can include a touch screen 1231 arranged on the front of the terminal device 110, which can collect touch operations on or near it, such as clicking buttons, dragging scroll boxes, etc.

[0275] Among them, the touch screen 1231 can be overlaid on the display screen 1232, or the touch screen 1231 and the display screen 1232 can be integrated to realize the input and output functions of the terminal device 110. After integration, it can be simply referred to as a touch display screen. The display unit 1230 in the present application can display application programs and corresponding operation steps.

[0276] The camera 1240 can be used to capture still images, and the object can publish comments through the application by taking pictures with the camera 1240. The camera 1240 can be one or more. The object generates an optical image through a lens and projects it onto a photosensitive element. The photosensitive element can be a charge-coupled device or a complementary metal-oxide-semiconductor phototransistor. The photosensitive element converts the optical signal into an electrical signal, and then transmits the electrical signal to the processor 1280 to convert it into a digital image signal.

[0277] The terminal device can also include at least one sensor 1250, such as an acceleration sensor 1251, a distance sensor 1252, a fingerprint sensor 1253, a temperature sensor 1254. The terminal device can also be configured with a gyroscope, a barometer, a hygrometer, a thermometer, an infrared sensor, a light sensor, a motion sensor, and other sensors.

[0278] ​The audio circuit 1260, the speaker 1261, and the microphone 1262 can provide an audio interface between the user and the terminal device 110. The audio circuit 1260 can convert the received audio data into an electrical signal, transmit the electrical signal to the speaker 1261, and convert the electrical signal into a sound signal output by the speaker 1261. The terminal device 110 can also be configured with a volume button for adjusting the volume of the sound signal. On the other hand, the microphone 1262 converts the collected sound signal into an electrical signal, which is received by the audio circuit 1260 and converted into audio data, and then the audio data is output to the communication component 1210 for transmission to another terminal device 110, for example, or to the memory 1220 for further processing.

[0279] The Bluetooth module 1270 is used to interact with other Bluetooth devices having a Bluetooth module through a Bluetooth protocol. For example, the terminal device can establish a Bluetooth connection with a wearable electronic device (e.g., a smart watch) also having a Bluetooth module through the Bluetooth module 1270, and then interact with the wearable electronic device.

[0280] The processor 1280 is the control center of the terminal device, which connects all parts of the terminal device through various interfaces and lines, and performs various functions of the terminal device and processes data by running or executing software programs stored in the memory 1220 and calling data stored in the memory 1220. In some embodiments, the processor 1280 can include one or more processing units; the processor 1280 can also integrate an application processor and a baseband processor, wherein the application processor mainly processes the operating system, the object interface, and the application program, and the baseband processor mainly processes wireless communication. It can be understood that the above-mentioned baseband processor can also not be integrated into the processor 1280. In the present application, the processor 1280 can run the operating system, the application program, the object interface display and the touch response, and the physiological data analysis method of the embodiments of the present application. In addition, the processor 1280 is coupled with the display unit 1230.

[0281] In some possible implementations, various aspects of the physiological data analysis method provided by the present application can also be implemented in the form of a program product, which includes a computer program for causing an electronic device to perform the steps of the physiological data analysis method according to various exemplary embodiments of the present application described above in the specification when the program product is run on the electronic device, for example, the electronic device can perform the steps as shown in Figure 3 .

[0282] The program product of the embodiments of the present application can employ any combination of one or more computer readable media. The computer readable media can be a computer readable signal medium or a computer readable storage medium. The computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the computer readable storage medium include an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0283] The program product of the embodiments of the present application can employ a portable compact disc read-only memory and include a computer program and can be executed on an electronic device. However, the program product of the present application is not limited thereto, and in the present document, the computer readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with a command execution system, apparatus, or device.

[0284] The computer readable signal medium can include a computer readable data signal embodied in a carrier wave, or the like, having a program embodied therein. The computer readable signal medium can also be any computer readable medium that is not a storage medium and that can communicate with a command execution system, apparatus, or device.

[0285] The computer program embodied on the computer readable medium can be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber cable, RF, and the like, or any suitable combination of the foregoing.

[0286] The computer program for performing the operations of the present application can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C++, and the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The computer program can execute entirely on the user's computing device, partly on the user's computing device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device or entirely on the remote computing device or server. In the latter scenario, the remote computing device can be connected to the user's computing device through any type of network, including a local area network or a wide area network, or the connection can be made to an external computing device.

[0287] Those skilled in the art will appreciate that embodiments of the present application can be devised for a method, a system, or a computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer readable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer readable program code thereon for use by or in connection with an instruction execution system. For the purposes of this description, a computer usable or computer readable storage medium can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.

[0288] Although preferred embodiments of the application have been described herein, those skilled in the art will appreciate that other changes and modifications can be made to the embodiments described without departing from the spirit and scope of the application. It is therefore intended that the appended claims encompass all such changes and modifications as fall within the scope of the application.

[0289] Obviously, numerous modifications and variations of the present application are possible in light of the above teachings. It is therefore to be understood that within the scope of the appended claims and their equivalents, the application can be practiced otherwise than as specifically described.

Claims

1. A method of analyzing physiological data, characterized by, The method comprises: obtaining original state information of an object to be analyzed and at least one type of object attribute information related to the object to be analyzed; wherein the original state information contains physiological data of the object to be analyzed in the environment in which the object to be analyzed is located, and the physiological data includes symptom information; obtaining a conditional probability of each preset state information under each preset physiological data analysis result respectively, and determining a first weight matrix for feature extraction based on the obtained conditional probability; performing feature extraction on the original state information based on the first weight matrix to obtain a corresponding first intermediate result; in a case where the preset state information is used to represent different symptom information preset and the preset physiological data analysis result is used to represent different disease results, the conditional probability represents a probability of occurrence of one symptom information under the condition of occurrence of one disease result; for each type of object attribute information in the at least one type of object attribute information, the following operations are performed respectively: performing feature extraction on one type of object attribute information based on a matching correlation degree between each preset physiological data analysis result and the one type of object attribute information to obtain a corresponding second intermediate result of the one type of object attribute information; wherein the feature extraction on the one type of object attribute information is realized by a second weight matrix based on the matching correlation degree; determining a class probability of each preset physiological data analysis result corresponding to the object to be analyzed based on the first intermediate result and at least one second intermediate result, and determining a target physiological data analysis result of the object to be analyzed based on each class probability.

2. The method of claim 1, wherein, The feature extraction on the one type of object attribute information based on the matching correlation degree between each preset physiological data analysis result and the one type of object attribute information to obtain a corresponding second intermediate result of the one type of object attribute information comprises: for one type of object attribute information, determining a second weight matrix for feature extraction on the one type of object attribute information based on a matching correlation degree between each preset physiological data analysis result and the one type of object attribute information; performing feature extraction on the one type of object attribute information based on the second weight matrix to obtain a corresponding second intermediate result.

3. The method of claim 1 or 2, wherein, The feature extraction on the original state information based on the first weight matrix to obtain a corresponding first intermediate result comprises: inputting the original state information of the object to be analyzed into a first feature network in a trained target analysis model; performing feature extraction on the original state information based on the first feature network to obtain the first intermediate result, wherein the first weight matrix in the first feature network is determined based on the conditional probability.

4. The method of claim 3, wherein, The feature extraction on the one type of object attribute information to obtain a corresponding second intermediate result of the one type of object attribute information comprises: inputting the one type of object attribute information of the object to be analyzed into a corresponding second feature network in the target analysis model, wherein the second feature network corresponds to the one type of object attribute information; Based on the second feature network corresponding to the object attribute information of the first type, the object attribute information corresponding to the second feature network is extracted to obtain a second intermediate result.

5. The method of claim 3, wherein, The target analysis model is trained by the following method: Obtain a sample data set, each sample data in the sample data set includes a sample object, original state information of the sample object, at least one type of object attribute information, and a physiological data analysis result label of the sample object; According to the sample data set, the analysis model to be trained is trained by cyclic iteration, and a corresponding target analysis model is output; wherein, in one cycle iteration process, the following operations are performed: Input the selected sample data into the analysis model to be trained to obtain the class probability of each type of preset physiological data analysis result corresponding to the sample object; Using the target loss function constructed based on the class probability of each type of preset physiological data analysis result corresponding to the sample object and the physiological data analysis result label, the parameters of the analysis model to be trained are adjusted.

6. The method of claim 5, wherein, In the analysis model to be trained, the first weight matrix in the feature network corresponding to the original state information is initialized based on the conditional probability; the second weight matrix in the feature network corresponding to each of the at least one type of object attribute information is initialized based on the matching correlation degree between each preset physiological data analysis result and itself; Then, the target loss function constructed based on the class probability of each type of preset physiological data analysis result corresponding to the sample object and the physiological data analysis result label is used to adjust the parameters of the analysis model to be trained, including: Based on the target loss function, the first weight matrix and each second weight matrix are adjusted.

7. The method of claim 5, wherein, The sample data set is obtained by: Divide the initial sample data into at least one sample data subset according to the preset physiological data analysis result category, and determine the sample quantity in the sample data subset corresponding to each preset physiological data analysis result category; The sample data in the sample data subset whose sample quantity does not reach the preset quantity is up-sampled to obtain at least one up-sampled sample data; Based on the up-sampled sample data and the initial sample data, the sample data set is constructed.

8. The method of claim 1 or 2, wherein, Based on the first intermediate result and at least one second intermediate result, the class probability of each type of preset physiological data analysis result corresponding to the object to be analyzed is determined, including: Determine the summation result obtained by accumulating and summing the first intermediate result and the at least one second intermediate result; The summation result is normalized to obtain the class probability of each type of preset physiological data analysis result corresponding to the object to be analyzed.

9. The method of claim 1 or 2, wherein, The object attribute information includes at least one of the following: object age information, object visit time information, and object gender information.

10. The method of claim 1 or 2, wherein, Based on each class probability, the target physiological data analysis result of the object to be analyzed is determined, including: The preset physiological data analysis result corresponding to the class probability reaching the reference threshold in each class probability is taken as the target physiological data analysis result of the object to be analyzed; or The category probabilities are sorted, and the preset physiological data analysis result corresponding to the category probability in the specified order range is taken as the target physiological data analysis result of the object to be analyzed.

11. An analysis device of physiological data, characterized in that The method comprises the following steps: An information acquisition unit is configured to acquire original state information of an object to be analyzed and at least one type of object attribute information related to the object to be analyzed; wherein the original state information comprises physiological data of the object to be analyzed in an environment in which the object to be analyzed is located, and the physiological data comprises symptom information; A feature extraction unit is configured to acquire a conditional probability of each preset state information under each preset physiological data analysis result respectively, and determine a first weight matrix used for feature extraction based on the acquired conditional probabilities; the original state information is subjected to feature extraction based on the first weight matrix to obtain a corresponding first intermediate result; in a case where the preset state information is used to represent different symptom information pre-set, and the preset physiological data analysis result is used to represent different disease results pre-set, the conditional probability represents a probability of occurrence of one symptom information under a condition of occurrence of one disease result; the feature extraction unit is further configured to perform the following operations on each type of object attribute information in the at least one type of object attribute information respectively: based on a matching and associating degree between each preset physiological data analysis result and one type of object attribute information, the one type of object attribute information is subjected to feature extraction to obtain a corresponding second intermediate result of the one type of object attribute information; wherein the feature extraction of the one type of object attribute information is realized through a second weight matrix based on the matching and associating degree; A result analysis unit is configured to determine a category probability of each preset physiological data analysis result corresponding to the object to be analyzed based on the first intermediate result and at least one second intermediate result, and determine a target physiological data analysis result of the object to be analyzed based on each category probability.

12. An electronic device, comprising: The method comprises the following steps:

13. A computer-readable storage medium, characterized in that, The method comprises the following steps:

14. A computer program product, characterised in that, The method comprises the following steps: The method comprises the following steps:

Images