Health state prediction network, model training and prediction method, device and system

Abstract

The invention provides a health state prediction network and a health state prediction model training method and device, and relates to the technical field of artificial intelligence and the technical field of education, in particular to the technical field of deep learning and the like. The specific implementation scheme of the health state prediction network is as follows: a long short-term memory network is used for obtaining a hidden state vector sequence according to behavior evaluation time sequence data of a detected object; the attention mechanism module is used for obtaining a time step weight sequence of the hidden state vector sequence and obtaining context information based on the hidden state vector sequence and the time step weight sequence; and the output layer is used for obtaining health state sequences of the tested object at different time points according to the context information.

Description

Technical Field

[0001] The present disclosure belongs to the field of computer technology, relates to the field of artificial intelligence technology, the field of educational technology, and specifically to the field of deep learning technology, in particular to a health status prediction network, a health status prediction model training method and device, a health status prediction method and device, a health status prediction system, an electronic device, and a computer-readable storage medium. Background Art

[0002] The pace of life in modern society is making more and more people feel great pressure. The pursuit of success, the pressure of study and work have become the main source of people's psychological burden. The increasing pressure makes people more likely to encounter various intellectual and physical health challenges. A study evaluated the impact of stress on health and found that stress may have a serious impact on diseases such as asthma, rheumatoid arthritis, anxiety, depression, cardiovascular disease, chronic pain, HIV / AIDS, stroke, cancer, etc., and may also accelerate the aging process and shorten life. In addition, stress may also lead to a decrease in brain cells and a decrease in brain volume. In addition to the long-term negative impact on physical health, stress may also cause serious mental distress and even lead to suicidal behavior. A survey of 67,000 teenagers showed that 75% of teenagers said they had experienced stress in the past year, and 20% of them said they felt stress more than five times. Studies have shown that teenagers are becoming more and more anxious. In the face of the general increase in stress, there is no way out. For example, taking activities such as meditation, exercise, getting enough sleep, reducing the frequency of checking emails, practicing yoga, listening to music, receiving massages and chewing gum are all effective ways to relieve stress. These measures can help reduce the damage caused by stress.

[0003] Formal education means going to school, exploring and learning new things. Teenagers’ lives can be difficult because of their complexity and constant challenges. However, it is entirely possible for teenagers to succeed in such an environment. Currently, many teenagers are complaining that they are under a lot of pressure in school. As the end of the semester approaches, the pressure is rising. Many teenagers feel more anxious and depressed at the end of the semester than at the beginning of the semester. The pressure of tests, homework and various exams in the learning process makes teenagers’ tension escalate.

[0004] In addition, there are other factors that affect adolescents' mental health, such as family conflicts, unclear future career plans and financial pressures. At the same time, trying to find a balance between school life and other areas of life may also put adolescents at risk of mental health. Those adolescents who showed mental health problems said that they did not get the necessary help and did not seek ways to solve emotional problems.

[0005] Therefore, how to provide appropriate support and help for the mental health of adolescents is an urgent problem that needs to be solved. Summary of the invention

[0006] The present disclosure provides a health status prediction network, a health status prediction model training method and device, an electronic device, and a computer-readable storage medium.

[0007] According to a first aspect, a health status prediction network is provided, which includes: a long short-term memory network, used to obtain a hidden state vector sequence based on the behavior evaluation time series data of the object under test; an attention mechanism module, used to obtain a time step weight sequence of the hidden state vector sequence, and obtain context information based on the hidden state vector sequence and the time step weight sequence; an output layer, used to obtain a health status sequence of the object under test at different time points based on the context information.

[0008] According to the second aspect, a method for training a health status prediction model is provided, the method comprising: obtaining a training data set; obtaining an initial health status prediction network, the health status prediction network adopting a health status prediction network as described in any implementation method of the first aspect; based on a health status sequence, calculating a network loss value of the health status prediction network; based on the network loss value, training the health status prediction network to obtain a trained health status prediction model.

[0009] According to the third aspect, a health status prediction method is provided, which includes: obtaining test data based on the acquired behavior evaluation time series data of the object under test; inputting the test data into a health status prediction model to obtain a health status sequence output by the health status prediction model, wherein the health status prediction model is trained using a method such as any implementation method of the second aspect.

[0010] According to a fourth aspect, a health status prediction model training device is provided, which includes: a sample acquisition unit, configured to acquire a training data set; a network acquisition unit, configured to acquire an initial health status prediction network, the health status prediction network adopts the health status prediction network described in any implementation method of the first aspect; a calculation unit, configured to calculate a network loss value of the health status prediction network based on a health status sequence; and a training unit, configured to train the health status prediction network based on the network loss value to obtain a trained health status prediction model.

[0011] According to the fifth aspect, a health status prediction device is provided, which includes: a data acquisition unit, configured to obtain test data based on the acquired behavior evaluation time series data of the object under test; an input unit, configured to input the test data into a health status prediction model to obtain a health status sequence output by the health status prediction model, wherein the health status prediction model is obtained by training using the device of the fourth aspect.

[0012] According to the sixth aspect, a health status prediction system is provided, which includes: a data acquisition terminal, used to obtain object information of the object under test, and based on the object information, obtain behavior evaluation time series data of the object under test; a prediction server, used to implement the health status prediction method described in any implementation method of the third aspect.

[0013] According to the seventh aspect, an electronic device is provided, which includes: at least one processor; and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method described in any implementation of the second aspect or the third aspect.

[0014] According to an eighth aspect, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to cause a computer to execute the method described in any implementation of the second aspect or the third aspect.

[0015] The health status prediction network provided by the embodiment of the present disclosure includes: a long short-term memory network, an attention mechanism module and an output layer. The long short-term memory network obtains a hidden state vector sequence according to the behavior evaluation time series data of the object under test; the attention mechanism module obtains a time step weight sequence of the hidden state vector sequence, and obtains context information based on the hidden state vector sequence and the time step weight sequence; the output layer obtains the health status sequence of the object under test at different time points according to the context information. Therefore, through the attention mechanism module in the health status prediction network, the most critical context information is effectively screened and deeply analyzed from the data, and the health status sequence is obtained based on the context information and the hidden state vector, thereby significantly improving the accuracy of the health status sequence prediction.

[0016] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it intended to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings are used to better understand the present solution and do not constitute a limitation of the present disclosure.

[0018] Figure 1 is a schematic diagram of the structure of an embodiment of a health status prediction network according to the present disclosure;

[0019] Figure 2 It is a structural diagram of the specific operation of the attention mechanism module in the present disclosure;

[0020] Figure 3 is a flowchart of an embodiment of a health status prediction model training method disclosed herein;

[0021] Figure 4 is a flow chart of an embodiment of a health status prediction method according to the present disclosure;

[0022] Figure 5 It is a structural schematic diagram of a health state sequence obtained by the health state prediction method disclosed in the present invention;

[0023] Figure 6 is a structural schematic diagram of an embodiment of a health status prediction model training device disclosed in the present invention;

[0024] Figure 7 is a structural schematic diagram of an embodiment of a health status prediction device according to the present disclosure;

[0025] Figure 8 is a structural schematic diagram of an embodiment of a health status prediction system according to the present disclosure;

[0026] Fig. 9 It is a block diagram of an electronic device used to implement the health status prediction model training method or health status prediction method of the embodiment of the present disclosure. DETAILED DESCRIPTION

[0027] Unless explicitly stated otherwise, throughout the specification and claims, the term “comprise” or variations such as “include” or “comprising”, etc., will be understood to include the stated elements or components but not to exclude other elements or components.

[0028] The technical solution of the present disclosure is described below by means of specific embodiments. It should be understood that one or more steps mentioned in the present disclosure do not exclude the existence of other methods and steps before and after the combination step, or other methods and steps may be inserted between these explicitly mentioned steps. It should also be understood that these examples are only used to illustrate the present disclosure and are not used to limit the scope of the present disclosure. Unless otherwise specified, the numbering of each method step is only for the purpose of identifying each method step, and does not limit the order of arrangement of each method or limit the scope of implementation of the present disclosure. Changes or adjustments in their relative relationships can also be regarded as the scope of implementation of the present disclosure without substantial changes in the technical content.

[0029] The sources of the raw materials and instruments used in the examples are not particularly limited and can be purchased from the market or prepared according to conventional methods known to those skilled in the art.

[0030] Limitations of traditional statistical models and basic machine learning algorithms: Existing happiness prediction methods mainly rely on traditional statistical models or basic machine learning algorithms such as linear regression, logistic regression, and support vector machine (SVM). When dealing with complex time series data, these methods are difficult to fully capture the potential nonlinear relationships and long-term dependencies in the data, resulting in low accuracy and stability of the prediction results.

[0031] Low prediction accuracy: Traditional methods often fail to achieve ideal accuracy in happiness prediction tasks, limiting their effectiveness and reliability in practical applications. For example, the prediction accuracy of existing models is usually far below 90%, which cannot meet the needs of early intervention and precise management.

[0032] In view of the defects in traditional technologies, this paper proposes a health status prediction network. Figure 1 A structural diagram 100 of an embodiment of a health status prediction network according to the present disclosure is shown. The health status prediction network includes: a long short-term memory network 101, an attention mechanism module 102 and an output layer 103.

[0033] Among them, the long short-term memory network 101 is used to obtain a hidden state vector sequence according to the behavior evaluation time series data of the object under test. The attention mechanism module 102 is used to obtain a time step weight sequence of the hidden state vector sequence, and obtain context information based on the hidden state vector sequence and the time step weight sequence. The output layer 103 is used to obtain a health state sequence of the object under test at different time points according to the context information.

[0034] In this embodiment, the long short-term memory network (LSTM) is a time recurrent neural network, which is specially designed to solve the long-term dependency problem of general RNN (recurrent neural network). All RNNs have a chain form of a repeated neural network module. The hidden state vector is the natural output of the internal state of the model, and is the implicit memory unit state when the LSTM processes sequence data at each time step. These hidden state vectors contain rich sequence information and are the key to LSTM capturing long-term dependencies. Therefore, the hidden state vector is not difficult to obtain, but is the output naturally generated by the LSTM model during training and reasoning.

[0035] In this embodiment, the attention mechanism module 102 is introduced into the traditional LSTM network as an additional component, and does not change the model structure of the LSTM itself. The core structure of the LSTM network, including the forget gate, input gate, output gate and cell state update mechanism in its unit, remains unchanged. The role of the attention mechanism module 102 is to weight these hidden state vectors after LSTM processes the sequence data and generates hidden state vectors to highlight the influence of key time steps in the sequence. The time step weight sequence generated by the attention mechanism module 102 allows the health status prediction network to pay more attention to those time steps that are more important to the health status sequence prediction when making the final prediction.

[0036] Specifically, the LSTM network generates a series of hidden state vectors h1, h2, ..., hN, i.e., a hidden state vector sequence, and the hidden state vectors in the hidden state vector sequence capture the time dependency in the behavior evaluation time series data. The attention mechanism module 102 generates a weighted context information by weighting the hidden state vectors in the hidden state vector sequence, and the context information is used for the final health state sequence prediction.

[0037] In this embodiment, the object under test is an object detected by a health status prediction network. The health status prediction network is used to predict the behavior evaluation time series data of the object under test to obtain the health status sequence of the object under test. This can effectively determine the historical and current health status of the object under test, thereby improving the accuracy and comprehensiveness of monitoring or diagnosis of the object under test.

[0038] In this embodiment, the behavior evaluation time series data is the behavior data of the measured object in the behavior activity and the performance data of the measured object due to the subordinate activity. For example, if the activity is a classroom activity, the behavior data is the action data of the measured object in different time periods in the classroom activity, and the performance data is the emotional feedback of the measured object on the classroom activity, academic performance, etc. When the measured object is a teenager and the health status time series is the teenager's happiness sequence, the behavior evaluation time series data includes not only academic performance, social media dynamics and other data, but also terminal device sensor data, such as exercise volume, sleep quality, etc. These behavior evaluation time series data can reflect the psychological and physiological state of teenagers from multiple dimensions, and provide comprehensive data support for happiness prediction.

[0039] In this embodiment, the time step refers to the time interval between each data point and its previous data point in the time series data. In time series analysis and prediction, the time step is an important concept that can be used to understand the time characteristics of data, build models and make predictions. The time step weight sequence includes at least one time step weight, each time step weight corresponds to a hidden state vector, and the long short-term memory network assigns a time step to each hidden state vector. The size of the time step of each hidden state vector can be the same or different. The time step weight of the hidden state vector can effectively reflect the key time step and characteristics of the hidden state vector in the behavior evaluation time series data.

[0040] In this embodiment, the hidden state vector sequence and the time step weight sequence correspond one to one. The above-mentioned obtaining of context information based on the hidden state vector sequence and the time step weight sequence includes: multiplying each hidden state vector with the corresponding time step weight and adding them to obtain the context information.

[0041] In this embodiment, the attention mechanism module 102 includes: a fully connected layer, an activation function, and a softmax function. Among them, the attention mechanism module 102 calculates the time step weight of each time step through a fully connected layer (or attention layer). Specifically, we first pass the hidden state vector of each time step through a fully connected layer, then apply an activation function (such as tanh) for nonlinear transformation, and finally generate a time step weight sequence α through a softmax function. The formula is shown in formulas (1) and (2):

[0042] e_t = \text{tanh}(W_a h_t + b_a) (1)

[0043] α= \frac{\exp(e_t)}{\sum_{t'} \exp(e_{t'})} (2)

[0044] In formula (1), W_a and b_a are trainable parameters.

[0045] In this embodiment, the output layer 103 may be a fully connected layer, and the hidden state vector is input into the fully connected layer (or softmax layer) for final classification or regression prediction.

[0046] In this embodiment, context information refers to the weighted representation of key time steps and features after being processed by the attention mechanism. The context information contains sufficient semantics and time dependencies to reflect the mental health status of the subject at different time points. Therefore, the output layer can predict the health status sequence of the subject at different time points based on this context information through a fully connected layer or other appropriate transformations.

[0047] The health status prediction network provided by the embodiment of the present disclosure includes: a long short-term memory network, an attention mechanism module and an output layer. The long short-term memory network obtains a hidden state vector sequence according to the behavior evaluation time series data of the object under test; the attention mechanism module obtains a time step weight sequence of the hidden state vector sequence, and obtains context information based on the hidden state vector sequence and the time step weight sequence; the output layer obtains the health status sequence of the object under test at different time points according to the context information. Therefore, through the attention mechanism module in the health status prediction network, the most critical context information is effectively screened and deeply analyzed from the data, and the health status sequence is obtained based on the context information and the hidden state vector, thereby significantly improving the accuracy of the health status sequence prediction.

[0048] In some embodiments of the present disclosure, the above-mentioned health status prediction network also includes: an attention sub-network, which is used to calculate the weight of the hidden state vector sequence in time steps to obtain the time step weight sequence of the hidden state vector sequence.

[0049] In this embodiment, the attention sub-network is a module independent of the long short-term memory network, the attention mechanism module and the output layer. The hidden state vector generated by the long short-term memory network is analyzed by the attention sub-network to generate the weights of each hidden state vector in the hidden state vector sequence at the time step, and obtain a time step weight sequence including the time step weight of at least one hidden state vector.

[0050] In this embodiment, the attention sub-network can dynamically adjust the weight of the time step according to the semantic information of the hidden state vector. This dynamic adjustment not only takes into account the relative importance of the time step, but also takes into account the semantic relevance between the hidden state vectors, thereby more accurately capturing the key information in the sequence data.

[0051] In this embodiment, when the behavior evaluation time series data is multi-modal and multi-source data, the attention sub-network can perform cross-modal weight allocation on hidden state vectors of different modalities, so that the model can better integrate information from different data sources and improve the accuracy of prediction.

[0052] The health status prediction network provided in this embodiment also includes an attention sub-network, which is used to calculate the weights of the hidden state vector sequence in time steps to obtain the time step weight sequence of the hidden state vector sequence, thereby providing a reliable implementation method for obtaining the time step weight sequence and improving the reliability and accuracy of the health status prediction network prediction.

[0053] In some optional implementations of the present disclosure, the above-mentioned attention mechanism module is also used to calculate the similarity between each hidden state vector in the hidden state vector sequence and a specific aspect feature to obtain a similarity value of each hidden state vector; and normalize the similarity value of each hidden state vector to obtain a time step weight of each hidden state vector.

[0054] In this optional implementation, specific aspect characteristics refer to characteristics that are related to the health status of the subject and are of great significance. For example, when the subject being measured is a teenager and the teenager's sense of happiness needs to be monitored, these specific aspect characteristics may include emotional tendencies, language patterns, fluctuations in academic performance, social behavior patterns, etc. These specific aspect characteristics can reflect the teenager's psychological state and level of happiness.

[0055] In this optional implementation, specific aspects of features can be obtained through the following ways: using external knowledge bases (such as sentiment dictionaries, psychological lexicons, etc.) to extract richer semantic features; using pre-trained language models (such as BERT, GPT, etc.) to extract deeper semantic features, which can better capture the emotional and psychological state information in the text when analyzing the happiness of teenagers; combining multiple modal data (such as text, images, voice, etc.) to extract multimodal features, which can more comprehensively reflect the psychological state of teenagers when analyzing the happiness of teenagers;

[0056] In this optional implementation, the attention mechanism module has a submodule specifically for generating time step weights, which calculates the similarity between each hidden state vector in the hidden state vector sequence and a specific aspect feature to obtain a similarity value of each hidden state vector; and normalizes the similarity value of each hidden state vector to obtain a time step weight of each hidden state vector.

[0057] like Figure 2 As shown, the specific operation of the attention mechanism module includes steps Step_1 to Step_4:

[0058] Step 1: The attention mechanism module obtains the hidden state vector sequence generated by the long short-term memory network

[0059] First, the LSTM network processes the input behavior evaluation time series data and generates a hidden state vector sequence h t (t is a value in 1-N), where t represents the time step.

[0060] Step 2: Calculate the time step weight sequence α

[0061] In order to determine the importance of each time step, the time step weight sequence α needs to be calculated. The attention mechanism module can generate the time step weight sequence α directly, or it can be done through an additional neural network (called the attention subnetwork), which will hide the state vector h t As input, it outputs the time step weight α. The time step weight α can be calculated by the softmax function to ensure that the sum of all weights is 1. Figure 2 In the submodule, each hidden state vector h t With specific aspects of characteristics V α The similarity calculation is performed to obtain the similarity value, and the similarity value is normalized to obtain the time step weight sequence α.

[0062] Step_3: Weighted summation

[0063] The time step weight α is combined with the hidden state vector h t Multiply them together and perform weighted summation on the hidden state vectors at all time steps to obtain a weighted hidden state vector sequence, namely the context information c. This weighted hidden state vector sequence represents the health status prediction network's evaluation of the behavior time series data (such as Figure 2 In the above figure, the attention representation of w1…wN) contains the most important information.

[0064] Step 4: Fusion of attention representation

[0065] The health state sequence y can be obtained directly through the context information, or the context information can be combined with the hidden state vector h of the LSTM network. t The final health status sequence y is obtained by fusion. This fusion can be performed in many ways, such as concatenation, summation or weighted summation. Figure 2 In the figure, the dotted box represents the hidden state vector sequence h output by the LSTM network. t .

[0066] The health status prediction network provided by the present invention, the attention mechanism module is also used to calculate the similarity between each hidden state vector in the hidden state vector sequence and the specific aspect feature to obtain the similarity value of each hidden state vector; the similarity value of each hidden state vector is normalized to obtain the time step weight of each hidden state vector, which can make the health status prediction network pay more attention to the specific aspect features, generate the time step weight of each hidden state vector based on the information of attention, and improve the reliability of obtaining the time step weight.

[0067] In some optional implementations of the present disclosure, the above-mentioned attention mechanism module is also used to collect semantic information of each hidden state vector in the hidden state vector sequence, and adjust the time step weight of each hidden state vector based on the semantic information of each hidden state vector.

[0068] In this optional implementation, the above-mentioned adjustment of the time step weight of each hidden state vector based on the semantic information of each hidden state vector includes: detecting whether the semantic information of each hidden state vector matches the preset health state, and if it matches, detecting whether the time step weight of the hidden state vector is greater than a preset upper limit threshold, and in response to being less than the preset upper limit threshold, adjusting the time step weight of the hidden state vector to the preset upper limit threshold.

[0069] The state prediction network provided by this optional implementation method, the attention mechanism module is also used to collect semantic information of each hidden state vector in the hidden state vector sequence, and adjust the time step weight of each hidden state vector based on the semantic information of each hidden state vector, thereby improving the flexibility and reliability of obtaining the time step weight.

[0070] In some optional implementations of the present disclosure, the above-mentioned health status prediction network also includes: a graph neural network module, which is used to collect social network data of the subject under test and extract key information from the social network data; and send the key information to a long short-term memory network so that the long short-term memory network generates a hidden state vector sequence based on the key information and behavior evaluation time series data.

[0071] In this optional implementation, the graph neural network module is used to collect the social network data of the subject under test, extract key information therein, and then send this key information to the long short-term memory network. In this case, the long short-term memory network combines this key information with the behavior evaluation time series data to generate a hidden state vector sequence. Among them, the social network data is the social group of the surrounding objects involved in the subject under test when performing the same activity as the behavior evaluation time series data.

[0072] In this optional implementation, after receiving the key information extracted by the graph neural network module, the long short-term memory network will use this information together with the behavior evaluation time series data as input to generate a new hidden state vector sequence. These hidden state vector sequences will comprehensively reflect the characteristics of the behavior data and social network data, thereby more comprehensively capturing the health status information of the subject.

[0073] In this optional implementation, the long short-term memory network generates a hidden state vector sequence based on "key information" and "behavior evaluation time series data" instead of just based on the behavior evaluation time series data. This combination enables the model to more accurately predict the health status sequence of the subject.

[0074] The health status prediction network provided in this embodiment also includes a graph neural network module, which is used to collect social network data of the object under test and extract key information from the social network data; the key information is sent to the long short-term memory network, so that the long short-term memory network generates a hidden state vector sequence based on the key information and the behavior evaluation time series data. To this end, the social network data and the behavior evaluation time series data are combined to jointly generate a hidden state vector sequence, which improves the comprehensiveness of the hidden state vector sequence information and improves the accuracy of data prediction by the health status prediction network.

[0075] This disclosure proposes a health status prediction model training method. Figure 3 A process 300 according to an embodiment of a health status prediction model training method disclosed herein is shown. The health status prediction model training method includes the following steps:

[0076] Step 301, obtaining a training data set.

[0077] In this embodiment, the execution subject on which the health status prediction model training method runs can obtain the training data set in a variety of ways. For example, the execution subject can obtain the training data set stored in the database server through a wired connection or a wireless connection. For another example, the user can obtain the training data set collected by the terminal by communicating with the terminal.

[0078] Here, the training data set may include at least one training data, which includes the behavior evaluation time series data of various objects and the annotated labels of the behavior evaluation time series data, and the annotated labels are used to characterize the health status of the object. When the object to be tested is a teenager, in order to realize the prediction of the happiness state of the teenager through the health status prediction network, the training data includes specific features and labels, which are selected and annotated according to the specific needs of the teenager's happiness prediction. For example, the data set may contain a large amount of information such as the behavior data, emotional feedback, and academic performance fluctuations of teenagers in the learning process, which are specially collected for the needs of this research.

[0079] In the technical solution of the present disclosure, the collection, storage, use, processing, transmission, provision and disclosure of the training data set involved are carried out after authorization and comply with relevant laws and regulations.

[0080] Step 302: Acquire an initial health status prediction network.

[0081] In this embodiment, the health status prediction network can adopt the health status prediction network disclosed in the above embodiment. For this purpose, the network structure of the health status prediction network can adopt any network structure of the health status prediction network in the embodiment disclosed in this disclosure. For example, the health status prediction network includes: a long short-term memory network, which is used to obtain a hidden state vector sequence based on the behavior evaluation time series data of the object under test; an attention mechanism module, which is used to obtain a time step weight sequence of the hidden state vector sequence, and obtain context information based on the hidden state vector sequence and the time step weight sequence; an output layer, which is used to obtain a health status sequence of the object under test at different time points based on the context information of the hidden state vector sequence.

[0082] In this embodiment, step 302 includes: defining the structure of the health status prediction network, including the number of layers of the health status prediction network, the number of hidden units in each layer, the attention mechanism module, and the implementation of the output layer. Initializing the weights and bias parameters of the initial health status prediction network. Usually, the weight parameters are initialized using a random initialization method (such as Xavier initialization), and the bias parameters are initialized to zero.

[0083] In step 303, the training data selected from the training data set is input into the health status prediction network to obtain a health status sequence output by the health status prediction network.

[0084] In this embodiment, the execution subject can select training data from the training data set obtained in step 301, and execute the training steps from step 303 to step 305 to complete an iterative training of the health status prediction model. Among them, the selection method and the number of selected training data from the training data set are not limited in this application, and the number of iterative training of the health status prediction model is not limited. For example, in one iterative training, multiple continuous training data can be randomly selected, and the network loss value of the health status prediction model is calculated through the selected training data, and the parameters of the health status prediction model are adjusted.

[0085] Step 304: Calculate the network loss value of the health state prediction network based on the health state sequence.

[0086] In this embodiment, during each iterative training of the health status prediction model, training data is selected from the training data set and input into the health status prediction model. Based on the health status sequence output by the health status prediction model, the network loss value of the health status prediction model is calculated.

[0087] In this embodiment, when calculating the network loss value, a cross entropy loss function can be used. The cross entropy loss function can measure the difference between two different probability distributions in the same random variable, which is expressed as the difference between the true probability distribution and the predicted probability distribution in machine learning. The smaller the value of the cross entropy loss function, the better the prediction effect of the health status prediction network.

[0088] In this embodiment, during the iterative training process of the health status prediction network, a gradient descent algorithm may be used to minimize the loss function of the health status prediction network, thereby iteratively optimizing the network parameters of the health status prediction network.

[0089] The original meaning of gradient is a vector, which means that the directional derivative of a loss function at that point reaches the maximum value along that direction, that is, the loss function changes fastest along that direction at that point and the rate of change is the largest. In deep learning, the main task of neural networks is to find the optimal network parameters (weights and biases) during learning. The optimal network parameters are also the parameters when the loss function is minimized.

[0090] Optionally, an appropriate optimizer may be selected to update the network parameters of the health status prediction network. In the present disclosure, since the Adam optimizer shows good performance in many deep learning tasks, the Adam optimizer may be used to optimize the network parameters of the health status prediction network.

[0091] Step 305: train the health status prediction network based on the network loss value to obtain a trained health status prediction model.

[0092] In this embodiment, the above training health status prediction network includes: inputting training data into the health status prediction network, calculating the health status sequence through the health status prediction network, i.e., forward propagation; calculating the network loss value according to the prediction result and the true label, i.e., calculating the loss; calculating the gradient of the network loss value with respect to the model parameters through the back propagation algorithm, i.e., back propagation; updating the model parameters using the optimizer according to the calculated gradient, i.e., parameter update. Repeat the above forward propagation, loss calculation, back propagation and parameter update process until the stop condition is met (such as reaching the predetermined number of training rounds, the loss on the validation set no longer decreases, etc.).

[0093] In this embodiment, the trained health status prediction model is obtained by inputting the selected training data into the health status prediction network for iterative training and adjusting the parameters of the health status prediction network to obtain the trained health status prediction model.

[0094] In this embodiment, the network loss value of the health status prediction network can be used to detect whether the health status prediction network meets the training completion condition. After the health status prediction network meets the training completion condition, a trained health status prediction model is obtained.

[0095] In this embodiment, the training completion condition includes: the network loss value of the health status prediction network is less than a first loss value threshold. The first loss threshold can be determined based on specific training requirements, for example, the first loss threshold is 0.01.

[0096] Optionally, in this embodiment, in response to the health status prediction network not meeting the training completion conditions, the relevant parameters in the health status prediction network are adjusted so that the network loss value of the health status prediction network converges, and the above training steps 303-305 are continued to be executed based on the adjusted health status prediction network.

[0097] The health status prediction model training method provided by the embodiment of the present disclosure first obtains a training data set; secondly, obtains an initial health status prediction network; then, inputs the training data selected from the training data set into the health status prediction network to obtain a health status sequence output by the health status prediction network; then, based on the health status sequence, calculates the network loss value of the health status prediction network; finally, based on the network loss value, trains the health status prediction network to obtain a trained health status prediction model; and provides a reliable implementation method for the training of the health status prediction model.

[0098] Before training the health status prediction model, it is necessary to prepare training data and validation data. Optionally, before calculating the network loss value of the health status prediction network based on the health status sequence, the above health status prediction model training method also includes: the specific steps are as follows: data enhancement can be performed on the training data set, such as data resampling, data balancing, etc., to improve the generalization ability of the model; the training data set is divided into a training set and a validation set (and a possible test set), wherein the training set is used to update the parameters of the health status prediction model, and the validation set is used to adjust the hyperparameters of the health status prediction model and prevent overfitting.

[0099] In some optional implementations of the present disclosure, the above method further includes: using at least one evaluation indicator of recall rate, F1 score, precision rate and accuracy rate to perform performance evaluation on the health status prediction model to obtain an evaluation result.

[0100] In this embodiment, the accuracy is calculated by the formula shown in formula (3). The accuracy is used to indicate the ratio of the number of samples correctly predicted by the health status prediction model to the total number of samples. As the most commonly used indicator for evaluating classification performance, the accuracy reflects the wide applicability and practicality of the health status prediction model.

[0101]

[0102] In formula (3), Tp (true positive) represents the number of samples that are actually positive (for example, the object under test is under stress) and correctly predicted as positive by the model; Tn (true negative) represents the number of samples that are actually negative (for example, the object under test is not under stress) and correctly predicted as negative by the model; Fp (false positive) represents the number of samples that are actually negative but mistakenly predicted as positive by the model; Fn (false negative) represents the number of samples that are actually positive but mistakenly predicted as negative by the model.

[0103] In this embodiment, the accuracy is calculated by the formula shown in formula (4). The accuracy is generally used to measure the degree of consistency between a series of effects. The accuracy is usually determined by comparing the difference between a set of actual results and the mathematical expectation of the set. The accuracy of the present disclosure refers to the time accuracy, which measures the consistency of the prediction results between different time steps. For example, if the model predicts that an object is in a stress state in multiple consecutive time steps, then we can say that the model has high time accuracy. This concept is reflected in formula (4).

[0104]

[0105] Usually, in classification tasks, Tp (True Positive) represents true positive examples, that is, the number of samples correctly predicted as positive by the model, and Fp (False Positive) represents false positive examples, that is, the number of samples that are actually negative but are incorrectly predicted as positive by the model. However, since Tp' and Fp' are used in formula (4), this represents the number of true positive examples and false negative examples after some adjustment or correction.

[0106] In this embodiment, the recall rate is calculated by the formula shown in formula (5), and the recall rate refers to the proportion of samples that are actually positive and are correctly predicted as positive by the model.

[0107]

[0108] Usually, in classification tasks, Tp (True Positive) represents true positive examples, that is, the number of samples correctly predicted as positive by the model, and Fn (False Negative) represents false negative examples, that is, the number of samples that are actually positive but are incorrectly predicted as negative by the health status prediction model. However, since Tp' and Fn' are used in formula (5), this represents the number of true positive examples and false negative examples after some adjustment or correction.

[0109] In this embodiment, the F1 score is used to measure the fit between the expected data and the prediction results of the health status prediction model.

[0110] This scoring criterion is described in detail in formula (6).

[0111]

[0112] In formula (6), Tpr (True Positive Rate) represents the proportion of samples predicted as positive by the health status prediction model among all samples that are actually positive; Fpr (False Positive Rate) represents the proportion of samples wrongly predicted as positive by the health status prediction model among all samples that are actually negative; Fnr (False Negative Rate) represents the proportion of samples wrongly predicted as negative by the health status prediction model among all samples that are actually positive.

[0113] The above indicators are often used to evaluate the performance of health status prediction models. Especially in unbalanced classification problems, the F1 score can provide more comprehensive information than the simple accuracy.

[0114] The method for performing performance evaluation on the health status prediction model provided by this optional implementation adopts multiple indicators including recall rate, F1 score, precision rate, and accuracy rate to perform performance evaluation on the health status prediction model, which can effectively test the performance of the health status prediction model and improve the testing efficiency of the health status prediction model.

[0115] Furthermore, based on the health status prediction model training method provided in the above embodiment, the present disclosure also provides an embodiment of a health status prediction method. The health status prediction method of the present disclosure combines artificial intelligence fields such as computer vision and deep learning. Figure 4 A process 400 according to an embodiment of a health status prediction method of the present disclosure is shown. The health status prediction method includes the following steps:

[0116] Step 401: based on the acquired behavior evaluation time series data of the tested object, obtain the data to be tested.

[0117] In this embodiment, the measured object is the object to be detected, and the measured object may be an adult, an employee, a student, etc. The behavior evaluation time series data of the measured object refers to the continuous behavior data of the measured object in the historical time period, and the behavior data may reflect the time series data of the current health status of the measured object.

[0118] In this embodiment, the data to be tested is the data obtained after preprocessing and feature engineering of the behavior evaluation time series data. The data to be tested can be directly input into the health status prediction model to obtain the health status sequence output by the health status prediction model.

[0119] In this embodiment, the execution subject of the health status prediction method can obtain the behavior evaluation time series data in a variety of ways. For example, the execution subject can obtain the behavior evaluation time series data stored in the database server through a wired connection or a wireless connection. For another example, the execution subject can also receive the behavior evaluation time series data collected in real time by a terminal or other device in real time.

[0120] Step 402: input the data to be tested into the health status prediction model to obtain the health status sequence output by the health status prediction model.

[0121] In this embodiment, the health status prediction model is trained using the above-mentioned health status prediction model training method. The health status prediction model uses an attention mechanism and a long short-term memory network to output the health status sequence of the object under test, thereby improving the accuracy of the health status sequence prediction.

[0122] In this embodiment, the health state sequence is a sequence that reflects the health state of the measured object. The health state sequence can be the mental health state or the physical health state of the measured object. The health state sequence output by the health state prediction model can provide a reference for doctors. The health state sequence includes: at least one health state, each health state can reflect the probability that the measured object is normal or abnormal.

[0123] In this embodiment, the execution subject can input the test data obtained from step 401 into the health status prediction model, thereby obtaining a health status sequence output by the health status prediction model. The health status sequence is used to indicate the current health status result of the tested object.

[0124] In this embodiment, the health status prediction model can be adopted as described above. Figure 3 The specific training process can be found in Figure 3 The relevant description of the embodiments will not be repeated here.

[0125] The health status prediction method provided in this embodiment obtains the data to be tested based on the behavior evaluation time series data of the object to be tested; the data to be tested is input into the health status prediction model to obtain the health status sequence output by the health status prediction model, thereby improving the reliability of the health status sequence of the object to be tested.

[0126] In some optional implementations of the present disclosure, the above-mentioned obtaining of the data to be tested based on the obtained behavior evaluation time series data of the tested object includes: obtaining the behavior content data and course learning data of the tested object, and using the behavior content data and course learning data as the behavior evaluation time series data; inputting the behavior evaluation time series data into the maximum value normalization formula to obtain the data to be tested.

[0127] In this optional implementation, the measured object can be a teenager, and the health state sequence output by the health state prediction model can be the happiness level value of the teenager; through the health state prediction method disclosed in this disclosure, the prediction ability of the mental health status of teenagers can be effectively improved. In order to ensure the standardization of the input data and lay a good foundation for its analysis, the study first pre-processes the data, and uses the maximum value normalization formula to achieve minimum-maximum normalization processing of the data to obtain a training data set. The behavior evaluation time series data is subjected to minimum-maximum normalization processing through the maximum value normalization formula, which can significantly improve the accuracy of the measured object's health state sequence.

[0128] In this optional implementation, the behavioral content data includes: relevant content information entered by the subject in the platform forum, which may be the content entered by the user when searching, posting or commenting, reflecting the user's interests, needs, emotional state, etc. For example, if a teenager frequently searches for keywords such as "stress management" and "emotional regulation", it may indicate that he or she is experiencing psychological pressure or emotional distress.

[0129] In this optional implementation, the course learning data is the data of the subject learning related labeled courses, where the labels may represent information such as the subject, content, or difficulty of the course. For example, the labels may include "mental health", "academic pressure", "emotional management", etc. By analyzing the labels of the courses that teenagers choose to study, we can understand their key areas of concern and thus infer their mental state.

[0130] In this optional implementation, the test data is input into a health status prediction model based on the attention mechanism, which can predict the happiness level of adolescents.

[0131] In this optional implementation, the transformed records have been normalized to reduce the burden on network calculations. The coordinate characters x, y, and z are normalized by min-max normalization to limit their value range to [0, 1]. The use of min-max normalization combined with learning bounded objectives can effectively improve the convergence performance of health status prediction models. Min-max normalization is a basic data preprocessing technique that ensures that numerical features or parameters can be adjusted to a specific range, usually 0 to 1. In order to improve the adaptability of data to analysis and target detection algorithms, evaluation time series data needs to be normalized through this system. By adjusting the data to the input range expected by the algorithm, min-max normalization significantly improves the efficiency and accuracy of the processing technique. In addition, by pushing outliers to the upper or lower limits of the normalized range, this method helps to highlight outliers and make them easier to distinguish from regular visitor patterns. The original data set is linearly transformed using the min-max normalization technique. When the minimum and maximum values ​​of a feature are normalized according to the minimum-maximum formula, the initial value of the feature will be replaced by a value in the interval [0, 1]. The specific maximum value normalization formula is shown in formula (7):

[0132]

[0133] Wherein, in formula (7), Min and Max represent the minimum value and maximum value of the test data X'' respectively. Accordingly, through minimum-maximum normalization, the initial value of the behavior evaluation time series data X'' is mapped into the test data X'' belonging to the interval [0, 1].

[0134] like Figure 5 As shown, after the behavior evaluation time series data is subjected to minimum-maximum normalization processing V, the test data is obtained, and the test data is input into the health status prediction model.

[0135] The method for obtaining the data to be tested provided in this embodiment obtains the behavior content data and course learning data of the object to be tested, and uses the behavior content data and course learning data as behavior evaluation time series data; the behavior evaluation time series data is input into the maximum value normalization formula to obtain the data to be tested. Thus, the behavior evaluation time series data is normalized by the normalization formula, thereby improving the accuracy of the data to be tested.

[0136] Optionally, after obtaining the health state sequence obtained by predicting the behavior evaluation time series data using the health state prediction model, Figure 5 As shown, the output result of the health status prediction model can also be evaluated by at least one evaluation index among the recall rate, F1 score, precision rate and accuracy rate to obtain an evaluation result.

[0137] Further references Figure 6As an implementation of the methods shown in the above figures, the present disclosure provides an embodiment of a health status prediction model training device. Figure 3 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0138] like Figure 6 As shown, the health status prediction model training device 600 provided in this embodiment includes: a sample acquisition unit 601, a network acquisition unit 602, a selection unit 603, a calculation unit 604, and a training unit 605. Among them, the above-mentioned sample acquisition unit 601 can be configured to acquire a training data set. The above-mentioned network acquisition unit 602 can be configured to acquire an initial health status prediction network, and the health status prediction network adopts the health status prediction network of the above-mentioned embodiment. The above-mentioned selection unit 603 can be configured to input the training data selected from the training data set into the health status prediction network to obtain a health status sequence output by the health status prediction network. The above-mentioned calculation unit 604 can be configured to calculate the network loss value of the health status prediction network based on the health status sequence. The above-mentioned training unit 605 can be configured to train the health status prediction network based on the network loss value to obtain a trained health status prediction model.

[0139] In this embodiment, the specific processing and technical effects of the sample acquisition unit 601, the network acquisition unit 602, the selection unit 603, the calculation unit 604, and the training unit 605 in the health status prediction model training device 600 can be referred to respectively. Figure 3 The relevant descriptions of step 301, step 302, step 303, step 304, and step 305 in the corresponding embodiment are not repeated here.

[0140] In some embodiments of the present disclosure, the above-mentioned device also includes an evaluation unit (not shown in the figure), and the above-mentioned evaluation unit is configured to: use at least one evaluation indicator of recall rate, F1 score, precision rate and accuracy rate to perform performance evaluation on the health status prediction model to obtain an evaluation result.

[0141] Further references Figure 7 As an implementation of the methods shown in the above figures, the present disclosure provides an embodiment of a health status prediction device. Figure 4 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0142] like Figure 7As shown, the health status prediction device 700 provided in this embodiment includes: a data obtaining unit 701, an input unit 702. The data obtaining unit 701 can be configured to obtain the data to be tested based on the acquired behavior evaluation time series data of the tested object. The input unit 702 can be configured to input the data to be tested into the health status prediction model to obtain a health status sequence output by the health status prediction model, wherein the health status prediction model is trained using the health status prediction model training device of the above embodiment. In this embodiment, in the health status prediction device 700: the specific processing of the data obtaining unit 701 and the input unit 702 and the technical effects brought about by them can be referred to respectively. Figure 4 The relevant descriptions of step 401 and step 402 in the corresponding embodiment are not repeated here.

[0143] In some embodiments of the present disclosure, the above-mentioned data acquisition unit 701 is configured to: obtain the behavior content data and course learning data of the object under test, and use the behavior content data and course learning data as behavior evaluation time series data; input the behavior evaluation time series data into the maximum value normalization formula to obtain the data to be tested.

[0144] Further references Figure 8 As an implementation of the methods shown in the above figures, the present disclosure provides an embodiment of a health status prediction system. Figure 4 The method embodiments shown correspond.

[0145] like Figure 8 As shown, the health status prediction system 800 provided in this embodiment includes: a data acquisition terminal 801 and a prediction server 802. The data acquisition terminal 801 is used to obtain the object information of the measured object, and based on the object information, obtain the behavior evaluation time series data of the measured object; the prediction server 802 is used to implement the health status prediction method described in the above embodiment.

[0146] In this embodiment, the object information of the measured object is the name and number of the measured object; an operation interface is set on the data acquisition terminal 801, and the user selects the measured object, such as teenagers, the elderly, etc., through the operation interface to obtain the object information of the measured object. Based on the selected measured object, the data acquisition terminal matches the behavior evaluation time series data related to the measured object from the database, and sends the behavior evaluation time series data to the prediction server 802.

[0147] In this embodiment, in the health status prediction system 800, the specific processing of the prediction server 802 and the technical effects thereof can be referred to in Figure 4 The relevant descriptions of step 401 and step 402 in the corresponding embodiment are not repeated here.

[0148] In the technical solution of the present disclosure, the collection, storage, use, processing, transmission, provision and disclosure of user personal information (such as the object information of the above-mentioned measured object) involved shall comply with the provisions of relevant laws and regulations and shall not violate public order and good morals.

[0149] In the technical solution of the present disclosure, the user personal information involved is also obtained after the user's authorization, and the user can be effectively analyzed through the user personal information.

[0150] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium and a computer program product.

[0151] Fig. 9 A schematic block diagram of an example electronic device 900 that can be used to implement an embodiment of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their modes are merely examples and are not intended to limit the implementation of the present disclosure described and / or required herein.

[0152] like Fig. 9 As shown, the device 900 includes a computing unit 901, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 902 or a computer program loaded from a storage unit 908 into a random access memory (RAM) 903. In the RAM 903, various programs and data required for the operation of the device 900 can also be stored. The computing unit 901, the ROM 902, and the RAM 903 are connected to each other via a bus 904. An input / output (I / O) interface 905 is also connected to the bus 904.

[0153] A number of components in the device 900 are connected to the I / O interface 905, including: an input unit 906, such as a keyboard, a mouse, etc.; an output unit 907, such as various types of displays, speakers, etc.; a storage unit 908, such as a disk, an optical disk, etc.; and a communication unit 909, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 909 allows the device 900 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0154] The computing unit 901 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 901 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 901 performs the various methods and processes described above, such as a health state prediction model training method or a health state prediction method. For example, in some embodiments, the health state prediction model training method or the health state prediction method may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit 908. In some embodiments, part or all of the computer program may be loaded and / or installed on the device 900 via the ROM 902 and / or the communication unit 909. When the computer program is loaded into the RAM 903 and executed by the computing unit 901, one or more steps of the health state prediction model training method or the health state prediction method described above may be performed. Alternatively, in other embodiments, the computing unit 901 may be configured to execute the health status prediction model training method or the health status prediction method in any other appropriate manner (eg, by means of firmware).

[0155] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0156] The program code for implementing the method of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable health state prediction model training device or a health state prediction device, so that when the program code is executed by the processor or controller, the mode / operation specified in the flow chart and / or block diagram is implemented. The program code may be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.

[0157] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, 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 disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0158] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0159] The systems and techniques described herein may be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0160] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps recorded in this disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved, and this document does not limit this.

[0161] The foregoing description of specific exemplary embodiments of the present disclosure is for the purpose of illustration and demonstration. These descriptions are not intended to limit the present disclosure to the precise form disclosed, and it is clear that many changes and variations can be made based on the above teachings. The purpose of selecting and describing the exemplary embodiments is to explain the specific principles of the present disclosure and its practical application, so that those skilled in the art can realize and utilize various different exemplary embodiments of the present disclosure and various different selections and changes. The scope of the present disclosure is intended to be defined by the claims and their equivalents.

Claims

1. A health status prediction network, the health status prediction network comprising: Long short-term memory network, used to evaluate the time series data based on the behavior of the object under test to obtain a hidden state vector sequence; An attention mechanism module, configured to obtain a time step weight sequence of the hidden state vector sequence, and obtain context information based on the hidden state vector sequence and the time step weight sequence; The output layer is used to obtain the health status sequence of the measured object at different time points according to the context information.

2. The health status prediction network according to claim 1, wherein: The health status prediction network also includes: The attention sub-network is used to calculate the weight of the hidden state vector sequence at the time step to obtain the time step weight sequence of the hidden state vector sequence.

3. The health status prediction network according to claim 1, wherein: The attention mechanism module is also used to calculate the similarity between each hidden state vector in the hidden state vector sequence and a specific aspect feature to obtain a similarity value of each hidden state vector; and normalize the similarity value of each hidden state vector to obtain a time step weight of each hidden state vector.

4. The health status prediction network according to claim 1, wherein: The attention mechanism module is also used to collect semantic information of each hidden state vector in the hidden state vector sequence, and adjust the time step weight of each hidden state vector based on the semantic information of each hidden state vector.

5. The health status prediction network according to any one of claims 1 to 4, wherein: The health status prediction network also includes: a graph neural network module, which is used to collect social network data of the object under test and extract key information from the social network data; send the key information to a long short-term memory network, so that the long short-term memory network generates a hidden state vector sequence based on the key information and the behavior evaluation time series data.

6. A health status prediction model training method, the method comprising: Get the training dataset; Acquire an initial health status prediction network, wherein the health status prediction network adopts the health status prediction network according to any one of claims 1 to 5; Inputting the training data selected from the training data set into the health status prediction network to obtain a health status sequence output by the health status prediction network; Based on the health state sequence, calculating a network loss value of the health state prediction network; Based on the network loss value, the health status prediction network is trained to obtain a trained health status prediction model.

7. The method according to claim 6, further comprising: At least one evaluation index among recall rate, F1 score, precision rate and accuracy rate is used to perform performance evaluation on the health status prediction model to obtain an evaluation result.

8. A health status prediction method, the method comprising: Based on the obtained behavior evaluation time series data of the tested object, the data to be tested is obtained; The data to be tested is input into a health status prediction model to obtain a health status sequence output by the health status prediction model, wherein the health status prediction model is trained using the method described in claim 6 or 7.

9. The method according to claim 8, wherein: The method of obtaining the test data based on the behavior evaluation time series data of the object under test includes: Acquire the behavior content data and course learning data of the subject to be measured, and use the behavior content data and the course learning data as behavior evaluation time series data; The behavior evaluation time series data is input into the maximum value normalization formula to obtain the data to be tested.

10. A health status prediction model training device, the device comprising: A sample acquisition unit is configured to acquire a training data set; A network acquisition unit, configured to acquire an initial health status prediction network, wherein the health status prediction network adopts the health status prediction network according to any one of claims 1 to 5; a selection unit configured to input the training data selected from the training data set into the health state prediction network to obtain a health state sequence output by the health state prediction network; a calculation unit configured to calculate a network loss value of the health state prediction network based on the health state sequence; The training unit is configured to train the health status prediction network based on the network loss value to obtain a trained health status prediction model.

11. A health status prediction device, comprising: A data obtaining unit is configured to obtain the data to be tested based on the obtained behavior evaluation time series data of the object under test; The input unit is configured to input the data to be tested into the health status prediction model to obtain a health status sequence output by the health status prediction model, wherein the health status prediction model is trained using the health status prediction model training device described in claim 10.

12. A health status prediction system, the health status prediction system comprising: A data acquisition terminal is used to obtain object information of the measured object, and based on the object information, obtain behavior evaluation time series data of the measured object; A prediction server is used to implement the health status prediction method described in claim 8 or 9 above.

13. An electronic device, characterized in that: include: at least one processor; and a memory communicatively coupled to the at least one processor; The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method according to any one of claims 1 to 9.

14. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 9.

Images