An optimization method for emotion prediction based on vital sign data

Abstract

The present invention relates to an emotion prediction optimization method based on vital sign data, which relates to the field of health management, including: activating an adaptive local predictor, collecting local vital sign data and inputting the adaptive local predictor, and outputting a first prediction result; if the first prediction result exceeds the emotion warning threshold and the number of consecutive overflows reaches a predetermined number limit, activating an adaptive global predictor, collecting global vital sign data and inputting the adaptive global predictor, and outputting a second prediction result as the emotion monitoring result of the target user. This application can solve the technical problem that traditional emotion prediction methods are difficult to take into account both the real-time and prediction accuracy of emotion prediction at the same time, and cannot meet the current emotion prediction needs. It can achieve rapid response and in-depth analysis under different emotional states, thereby effectively improving the accuracy and reliability of emotion prediction while ensuring real-time and efficiency, and saving computing power resource consumption.

Description

Technical Field

[0001] The present invention relates to the field of health management, and in particular to a mood prediction optimization method based on physical sign data. Background Art

[0002] Emotion prediction technology identifies and analyzes an individual's emotional state based on physiological and behavioral characteristics (such as heart rate, galvanic skin response, and respiratory rate). With the widespread adoption of smart wearable devices, health monitoring systems, and mental health applications, the demand for emotion prediction is growing. However, traditional emotion prediction methods typically rely on either simple prediction models to ensure real-time performance at the expense of accuracy, or rely on complex, large models to improve accuracy, resulting in slow system response and high computational overhead, making them unable to meet real-time monitoring requirements.

[0003] In summary, traditional emotion prediction methods are difficult to take into account both the real-time and prediction accuracy of emotion prediction, and cannot meet the current emotion prediction needs. Summary of the Invention

[0004] The present invention aims to solve the technical problem that traditional emotion prediction methods are difficult to take into account both the real-time performance and prediction accuracy of emotion prediction and cannot meet the current emotion prediction needs. It provides an emotion prediction optimization method based on vital sign data to solve the problem.

[0005] The technical solution of the present invention to solve the above technical problems is as follows:

[0006] In a first aspect, the present invention provides an emotion prediction optimization method based on vital sign data, comprising: building an emotion prediction plug-in matching table based on a predetermined basic parameter interval, a predetermined motion parameter interval, and a predetermined environmental parameter interval, wherein the emotion prediction plug-in includes a local emotion predictor and a global emotion predictor, the local emotion predictor being constructed based on local vital sign indicators, and the global emotion predictor being constructed based on global vital sign indicators; inputting the target user's basic data, expected motion data, and environmental data into the emotion prediction plug-in matching table to obtain an adapted local predictor and an adapted global predictor; activating the adapted local predictor, collecting local vital sign data according to the local vital sign indicators, inputting the adapted local predictor, and outputting a first prediction result; if the first prediction result exceeds an emotion warning threshold, and the number of consecutive overflows reaches a predetermined limit, activating the adapted global predictor, collecting global vital sign data according to the global vital sign indicators, and inputting the adapted global predictor, and outputting a second prediction result as the emotion monitoring result of the target user.

[0007] In a second aspect, the present invention further provides an electronic device, comprising:

[0008] At least one processor; a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor so as to enable the at least one processor to perform the steps of any one of the methods described in the first aspect above.

[0009] In a third aspect, a computer-readable storage medium is provided, wherein a computer program is stored on the computer-readable storage medium, and when the computer program is executed, the steps of the method described in any one of the first aspects are implemented.

[0010] The beneficial effects of the present invention are as follows: by building an emotion prediction plug-in matching table based on a predetermined basic parameter interval, a predetermined motion parameter interval and a predetermined environmental parameter interval, wherein the emotion prediction plug-in includes a local emotion predictor and a global emotion predictor, the local emotion predictor is constructed based on a local vital sign index, and the global emotion predictor is constructed based on a global vital sign index; the basic data, expected motion data and environmental data of the target user are input into the emotion prediction plug-in matching table to obtain an adapted local predictor and an adapted global predictor; the adapted local predictor is activated, local vital sign data is collected according to the local vital sign index and input into the adapted local predictor, and a first prediction result is output; if the If the first prediction result exceeds the emotion warning threshold and the number of consecutive overflows reaches a predetermined limit, the adaptive global predictor is activated, global vital sign data is collected according to the global vital sign indicators and input into the adaptive global predictor, and a second prediction result is output as the emotion monitoring result of the target user; that is, by adopting a local emotion prediction model for short-term and efficient monitoring, it is possible to quickly respond to emotional anomalies, and switch to a global emotion prediction model for precise analysis when the abnormal emotion persists, which can achieve rapid response and in-depth analysis under different emotional states, thereby effectively improving the accuracy and reliability of emotion prediction while ensuring real-time and efficiency, and saving computing power resource consumption. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Figure 1 A schematic flow chart of an emotion prediction optimization method based on vital sign data provided by the present invention;

[0012] Figure 2 A schematic structural diagram of the electronic device provided by the present invention;

[0013] Figure 3 A schematic structural diagram of a computer-readable storage medium provided by the present invention.

[0014] In the accompanying drawings, the components represented by the reference numerals are described as follows:

[0015] Electronic device 500 , memory 510 , processor 520 , first computer program 511 , computer-readable storage medium 600 , second computer program 611 . DETAILED DESCRIPTION

[0016] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative efforts shall fall within the scope of protection of the present invention.

[0017] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first" or "second" may explicitly or implicitly include one or more of the specified features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0018] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or illustration". Any embodiment of the present invention described as "for example" is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any person skilled in the art to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes are not elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed herein.

[0019] Example 1, as Figure 1 As shown, an embodiment of the present invention provides an emotion prediction optimization method based on vital sign data, which specifically includes the following steps:

[0020] An emotion prediction plug-in matching table is constructed based on a predetermined basic parameter interval, a predetermined motion parameter interval, and a predetermined environmental parameter interval, wherein the emotion prediction plug-in includes a local emotion predictor and a global emotion predictor, the local emotion predictor being constructed based on local vital sign indicators, and the global emotion predictor being constructed based on global vital sign indicators.

[0021] Furthermore, to construct a global emotion predictor, the present application also includes:

[0022] A vital sign indicator space is configured, wherein the vital sign indicator space includes several vital sign indicators, and the vital sign indicators include at least heart rate, skin electrical response, respiratory rate, blood oxygen saturation, body temperature, blood pressure, brain waves, etc.; collection convenience and emotion correlation analysis are performed on the several vital sign indicators respectively, and several collection convenience coefficients and several emotion correlation coefficients are obtained, and several indicator influences are obtained by weighted calculation; the several indicator influences are screened according to a predetermined influence threshold, and the vital sign indicators corresponding to the indicator influences greater than the predetermined influence threshold are selected to obtain a global vital sign indicator.

[0023] Specifically, in emotion prediction and health monitoring, configuring an appropriate physical sign indicator space is the basis for achieving efficient and accurate prediction. The physical sign indicator space includes multiple physiological and biological signals. These indicators work together to provide multi-dimensional information about the individual's emotional state. First, configure the physical sign indicator space, where the physical sign indicator space includes several individual sign indicators. The physical sign indicators include at least heart rate, skin electrical response, respiratory rate, blood oxygen saturation, body temperature, blood pressure, brain waves, etc. For example, heart rate reflects the number of heart beats per minute and is usually closely related to emotional changes. For example, in emotional states such as anxiety, fear or anger, the heart rate will accelerate; when relaxed, the heart rate will be slower.

[0024] Next, the aforementioned individual physical indices were analyzed for ease of collection and emotional relevance. Ease of collection refers to the ease of measurement, equipment requirements, and feasibility of real-time collection for each physical indices. Indicators with higher ease of collection typically can be monitored on low-cost, efficient equipment, such as smart bracelets or wearable devices. The ease of collection coefficient is used to measure this attribute. For example, heart rate monitoring is typically performed using photoplethysmography (PPG) technology, which is convenient and many smart devices support real-time monitoring, resulting in a higher ease coefficient. EEG monitoring requires specialized electrodes and equipment and is typically more complex. The ease coefficient is lower. Emotional relevance refers to the strength of the correlation between physical indices and emotional changes, quantified by the emotional relevance coefficient, which is typically derived from physiological studies or data analysis. For example, heart rate is closely related to emotional changes, especially in situations of anxiety, anger, and stress, where heart rate typically accelerates, resulting in a higher emotional relevance coefficient. Respiratory rate fluctuates significantly when stressed, anxious, or relaxed, and is particularly pronounced during emotional fluctuations, resulting in a moderate emotional relevance coefficient. Blood oxygen saturation is indirectly associated with emotional fluctuations, primarily by affecting changes in breathing patterns, resulting in a lower emotional relevance coefficient.

[0025] Then, weights are configured for the collection convenience coefficient and the emotion correlation coefficient, which can be adjusted according to needs to ensure a balance between convenience and emotion correlation in the calculation. Then, according to the configured weights, a weighted calculation is performed on several collection convenience coefficients and several emotion correlation coefficients to obtain several indicator influences, among which the indicators with high influence can be used as the focus of attention in the emotion prediction model. The influences of the several indicators are further screened according to a predetermined influence threshold. In order to screen out the most important global vital sign indicators, a predetermined influence threshold is required to filter out those indicators with low influence, ensuring that only vital sign indicators that have a significant contribution to emotion prediction are selected. An appropriate influence threshold can be selected based on actual application requirements, device resource limitations, and accuracy requirements. For example, the threshold can be set to 0.5, which means that vital sign indicators with an influence of less than 0.5 will be excluded. Next, the vital sign indicators corresponding to the indicator influences greater than the predetermined influence threshold are selected to obtain the global vital sign indicators. The global vital sign indicators refer to those indicators that have a significant contribution to emotion prediction, are easy to collect, and have a high correlation with the emotional state.

[0026] Based on the principle of ensemble learning, the global emotion predictor is constructed according to the global vital sign indicators.

[0027] Furthermore, the present application further comprises the following steps:

[0028] Randomly select any parameter in the predetermined basic parameter interval, the predetermined motion parameter interval and the predetermined environmental parameter interval for combination to obtain a first scene parameter; perform retrieval with the first scene parameter and the global vital sign index as constraints to obtain a sample global vital sign index set and a sample emotion coefficient set; use the sample global vital sign index set and the sample emotion coefficient set as training data to supervise N emotion prediction operators to obtain N global emotion prediction branches that meet convergence, wherein the emotion prediction operator includes at least a BP neural network, a support vector machine and a random decision forest, and N is an integer greater than or equal to 3; obtain N prediction accuracies of the N global emotion prediction branches, and set N credible weight coefficients according to the N prediction accuracies; based on the principle of ensemble learning, construct a first global emotion predictor according to the N global emotion prediction branches and the N credible weight coefficients, and establish a first mapping association between the first scene parameter and the first global emotion predictor.

[0029] Specifically, in the emotion prediction optimization method based on vital sign data, the predetermined basic parameter ranges, motion parameter ranges, and environmental parameter ranges are designed to personalize emotion prediction based on the user's individual characteristics, activity status, and environmental conditions. The acquisition and setting of these parameters helps improve the accuracy and adaptability of emotion prediction.

[0030] First, a predetermined basic parameter interval, a predetermined motion parameter interval, and a predetermined environmental parameter interval are obtained. Basic parameters refer to basic physical parameters such as age, gender, height, and weight. For example, the age interval is set to different intervals such as 0 to 18 years old, 19 to 35 years old, 36 to 60 years old, and over 60 years old. Motion parameters include exercise type and exercise intensity. Environmental parameters include ambient temperature. Next, any parameter in the predetermined basic parameter interval, predetermined motion parameter interval, and predetermined environmental parameter interval is randomly selected and combined to obtain the first scene parameters. For example, the first scene parameters include 19 to 35 years old, male, 175 cm, running, and room temperature of 25 to 30 degrees Celsius.

[0031] Next, information retrieval is performed with the first scene parameters and global vital signs indicators as constraints to screen out sample data that meets the requirements of emotion prediction. Information retrieval can be performed through big data technology to obtain a sample global vital signs indicator set and a sample emotion coefficient set, where the emotion coefficient can be calculated based on historical data, user feedback or emotion labels to represent the intensity of the emotional state. For example, 0 represents no emotion, 10 represents an extreme emotional state, the angry emotion coefficient can be set to a higher value (such as 8 or 9), and the calm emotion coefficient is a lower value (such as 2 or 3).

[0032] Then, obtain N emotion prediction operators, wherein the emotion prediction operators include at least BP neural network, support vector machine and random decision forest, N is an integer greater than or equal to 3, BP neural network is a commonly used feedforward neural network, which is trained using back propagation algorithm and can handle nonlinear relationships; SVM is a powerful classification and regression model, which is particularly suitable for classification tasks of high-dimensional data. SVM optimizes classification accuracy by finding the maximum margin hyperplane; random decision forest performs classification and regression through a voting mechanism of multiple decision trees, has strong robustness, and can handle high-noise data; technical personnel in this field can also select an adaptive prediction operator according to actual conditions; further use the sample global sign index set and the sample emotion coefficient set as training data, and divide the training data into N parts to obtain N training data sets; then, use the sample global sign index as input and the sample emotion coefficient as output, and use N data sets to supervise the training of N emotion prediction operators respectively. For example, construct a first global emotion prediction branch based on the BP neural network, and use the first training data set to supervise the training of the first global emotion prediction branch. First, initialize the structure and parameters of the BP neural network, including input The number of neurons in the forward layer, hidden layer and output layer is randomly initialized with weights and biases; then, the input (global vital sign indicator set) in the training data set is passed into the neural network, and calculations are performed through each layer to finally obtain the prediction result; then, based on the gap between the predicted value and the actual emotion coefficient, the loss function (such as mean square error MSE) is calculated, and the gradient of the loss function with respect to the network parameters (weights and biases) is calculated through the back propagation algorithm, and these parameters are updated, and the gradient descent algorithm (or other optimization algorithms) is used to minimize the loss function; the above process (forward propagation, loss calculation, back propagation, parameter update) is repeated until the loss function converges to the minimum value, or the preset maximum number of training rounds is reached; the first global emotion prediction branch that has completed training is output; and then N global emotion prediction branches that meet convergence are obtained in turn.

[0033] Further, N prediction accuracies of the N global emotion prediction branches are obtained, where prediction accuracy refers to the accuracy of the model output. Then, based on the prediction accuracy of each emotion prediction branch, a credible weight coefficient can be assigned to it. Branches with higher prediction accuracy are assigned a larger weight, and vice versa, branches with lower prediction accuracy are assigned a smaller weight, resulting in N credible weight coefficients. Finally, based on the principle of ensemble learning, a first global emotion predictor is constructed based on the N global emotion prediction branches and the N credible weight coefficients. That is, based on the principle of ensemble learning, the output results of the N emotion prediction branches are synthesized into a final prediction result using a weighted average method. A first mapping association is also established between the first scene parameter and the first global emotion predictor.

[0034] By building a global emotion predictor based on the principle of ensemble learning, that is, combining the output results of multiple prediction models for comprehensive prediction, the prediction bias of a single model can be reduced and the stability and accuracy of emotion prediction can be improved.

[0035] Furthermore, to construct a local emotion predictor, the present application also includes:

[0036] In a two-dimensional space, a correlation fluctuation curve between the global physical sign index and the emotional characteristics is constructed to obtain a plurality of global correlation fluctuation curves.

[0037] Specifically, in the emotion prediction optimization method, the relationship between global vital signs indicators (such as heart rate, skin galvanic response, respiratory rate, etc.) and emotional characteristics (such as emotional intensity, emotional labels, etc.) can be represented by correlation fluctuation curves. By constructing these curves, the dynamic correlation and change trend between vital sign data and emotional state can be revealed. Then, in a two-dimensional space, the correlation fluctuation curves of the global vital sign indicators and emotional characteristics are constructed based on the sample data to obtain multiple global correlation fluctuation curves. For example, in a two-dimensional space, the X-axis represents the global vital sign indicators (such as heart rate, skin galvanic response, etc.), each vital sign indicator has a corresponding numerical value, and the Y-axis represents the emotional characteristics (such as emotional intensity or emotional label). The emotional coefficient can be a continuous value (such as emotional intensity) or use emotional labels (such as happiness, anxiety, etc.). Based on the sample data, in chronological order, the correlation fluctuation curve is constructed. The key to this process is to correspond the vital sign data and emotional characteristics at each time point and connect them to form a fluctuation curve. By observing the fluctuation of the curve, the dynamic relationship between the vital sign data and emotional changes can be analyzed.

[0038] performing similarity dimensionality reduction on the multiple global correlation fluctuation curves to obtain multiple difference correlation fluctuation curves, setting the physical sign indicators corresponding to the multiple difference correlation fluctuation curves as difference physical sign indicators, and obtaining multiple difference physical sign indicators;

[0039] Furthermore, the present application further comprises the following steps:

[0040] A first associated fluctuation curve is randomly selected from the multiple global associated fluctuation curves, and similarity comparisons are performed on the first associated fluctuation curve and other associated fluctuation curves in the multiple global associated fluctuation curves to obtain multiple first similarities; the number of first similarities that are less than a predetermined similarity index is counted and set as a first difference; if the first difference is greater than the predetermined difference index, the first associated fluctuation curve is set as a difference associated fluctuation curve and added to the multiple difference associated fluctuation curves.

[0041] Specifically, any one of the multiple global correlation fluctuation curves is randomly selected and set as the first correlation fluctuation curve; then, a similarity comparison is performed on the first correlation fluctuation curve and the other correlation fluctuation curves in the multiple global correlation fluctuation curves. For example, the similarity comparison can be performed using Euclidean distance, which is a standard method for measuring the straight-line distance between two points. It is used to measure the difference between two curves at each corresponding time point. The closer the value is to 1, the closer the two curves are; and multiple first similarities are obtained. The number of first similarities that are less than a predetermined similarity index is further counted. The predetermined similarity threshold can be set according to the actual application scenario. For example, if the similarity is less than 0.7, it is considered to be a large difference. The quantitative statistical result is set as the first difference. The first difference represents the overall deviation of the first correlation fluctuation curve from the other correlation fluctuation curves in the multiple global correlation fluctuation curves. If the first difference is greater than a predetermined difference index (which can be set according to the number of indicators, such as 5), the first correlation fluctuation curve is set as a difference correlation fluctuation curve and added to the multiple difference correlation fluctuation curves.

[0042] Finally, the physical sign indicators corresponding to the multiple difference-correlated fluctuation curves are set as difference physical sign indicators to obtain multiple difference physical sign indicators. By similarity comparison based on Euclidean distance, the difference-correlated fluctuation curves are screened and the corresponding difference physical sign indicators are extracted, which can effectively identify physical sign data with significant differences during the process of emotional changes.

[0043] The plurality of difference physical sign indicators are sorted in descending order according to the degree of influence of the indicators, and the first preset number of indicators in the difference physical sign indicator sequence are selected as local physical sign indicators.

[0044] Specifically, the multiple differential vital sign indicators are further sorted by their influence from largest to smallest to obtain a differential vital sign indicator sequence. A preset number of indicators (e.g., the first three) from the differential vital sign indicator sequence are then selected as local vital sign indicators. By sorting the differential vital sign indicators by their influence and selecting the top N indicators as local vital sign indicators, the emotion prediction system can ensure that it provides highly real-time predictions while focusing on the most critical vital sign data. This approach combines short-term monitoring with high efficiency, effectively improving the overall performance of the system while ensuring the accuracy of emotion predictions.

[0045] Based on the BP neural network, a local emotion predictor is constructed according to the local physical sign indicators.

[0046] Furthermore, the present application further comprises the following steps:

[0047] A search is performed with the first scene parameter and local sign indicator as constraints to obtain a sample local sign indicator set and a sample emotion coefficient set; Q local prediction branches are configured based on the BP neural network, where Q is an integer greater than 1 and less than 5; the sample local sign indicator set and the sample emotion coefficient set are used as a sample data set, and the sample data set is divided into Q equal parts, and Q parts are selected with replacement Q times to construct a first training set, and Q training sets are obtained by iterative selection; the Q local prediction branches are supervised and trained using the Q training sets until convergence, a first local emotion predictor is constructed based on the trained Q local prediction branches, and a second mapping association between the first scene parameter and the first local emotion predictor is established.

[0048] Specifically, in the emotion prediction optimization method, the local emotion predictor is used to quickly predict emotions based on specific physical signs indicators; first, the first scene parameters and local physical signs indicators are used as constraints for retrieval to obtain a sample local physical signs indicator set and a sample emotion coefficient set; then, Q local prediction branches are configured based on the BP neural network, where Q is an integer greater than 1 and less than 5; then the sample local physical signs indicator set and the sample emotion coefficient set are used as a sample data set, and the sample data set is divided into Q equal parts to obtain Q data sets, and the Q data sets are selected Q times with replacement to construct a first training set, and the same method is used to iteratively select Q times to obtain Q training sets.

[0049] Finally, the Q local prediction branches are supervised and trained using the Q training sets, with the sample local sign indicators as input and the sample emotion coefficients as supervision. First, the BP neural network is used for training, and each local prediction branch learns independently. Each local prediction branch learns how to predict the target emotion coefficient based on the input local sign indicators. Then, a standard loss function (such as mean square error (MSE)) is used to evaluate the difference between the prediction results of each local prediction branch and the true label, and the gradient descent algorithm is used to adjust the network parameters until the network converges, that is, the value of the loss function approaches the minimum. When the loss function of each local prediction branch reaches a predetermined convergence standard (such as the error is less than a certain threshold, or the number of training iterations reaches a predetermined value), the training process stops. After the training is completed, Q local emotion prediction branches are obtained. Then, a first local emotion predictor is constructed based on the trained Q local prediction branches, and a second mapping association between the first scene parameter and the first local emotion predictor is established.

[0050] Furthermore, this application also includes:

[0051] Based on the first mapping association and the second mapping association, the emotion prediction plug-in matching table is constructed according to the first scene parameter, the first global emotion predictor, and the first local emotion predictor.

[0052] Specifically, based on the principle of classification decision-making, the first mapping association and the second mapping association establish a first matching relationship based on the first scene parameter, the first global emotion predictor, and the first local emotion predictor. This matching relationship can dynamically call an adapted emotion prediction plug-in (including a local emotion predictor and a global emotion predictor) based on the specific first scene parameter, ensuring that the emotion prediction system makes the most reasonable decision based on the user's real-time data and emotional state. Then, using the same method, multiple scene parameters and multiple global emotion predictors and multiple local emotion predictors corresponding to the multiple scene parameters are sequentially obtained, multiple matching relationships are established, and the emotion prediction plug-in matching table is constructed based on the multiple matching relationships.

[0053] By constructing an emotion prediction plug-in matching table based on the classification decision principle, we can intelligently select suitable emotion predictors according to the different characteristics of scene parameters. This method not only improves the efficiency and real-time performance of emotion prediction, but also optimizes the accuracy and reliability of emotion prediction by dynamically adapting different predictors.

[0054] The target user's basic data, expected motion data, and environmental data are input into the emotion prediction plug-in matching table to obtain an adapted local predictor and an adapted global predictor.

[0055] Specifically, the basic data (height, weight, etc.), expected motion data (to-be-performed motion) and environmental data (ambient temperature, etc.) of the target user are collected; and the basic data, expected motion data and environmental data are input into the emotion prediction plug-in matching table for matching to obtain an adapted local predictor and an adapted global predictor.

[0056] The adapted local predictor is activated, local sign data is collected according to the local sign index and input into the adapted local predictor, and a first prediction result is output.

[0057] Specifically, the adaptive local predictor is then activated. Local vital sign data of the target user is then collected based on the local vital sign indicators. This data is then input into any prediction branch of the adaptive local predictor for prediction. Initially, to conserve computing resources, only a single branch is used for prediction, outputting a first prediction result. By activating only a single prediction branch, system computing resource consumption is reduced, particularly during emotion prediction, enabling rapid initial prediction results without sacrificing real-time performance.

[0058] If the first prediction result exceeds the emotion warning threshold and the number of consecutive overflows reaches a predetermined limit, the adaptive global predictor is activated, global vital sign data is collected according to the global vital sign indicators and input into the adaptive global predictor, and a second prediction result is output as the emotion monitoring result of the target user.

[0059] Furthermore, this application also includes:

[0060] Obtain a first prediction result, wherein the first prediction result is a first predicted emotion coefficient; if the first predicted emotion coefficient is greater than the emotion warning threshold, calculate the emotion coefficient deviation, and set the ratio of the emotion coefficient deviation to the maximum emotion coefficient deviation as the emotion fluctuation scale; round the emotion fluctuation scale and multiply it by Q to obtain the number of calls of the local prediction branch, wherein the initial number of call branches of the adapted local predictor is 1; randomly select the local prediction branches of the said number of calls from the Q local prediction branches of the adapted local predictor, perform continuous monitoring, and record the number of consecutive overflows, wherein overflow refers to the predicted emotion coefficient being greater than the emotion warning threshold. During the continuous monitoring process, the number of local prediction branches will be iteratively updated according to the emotion coefficient deviation in each monitoring result; if the number of consecutive overflows reaches a predetermined limit, activate the adapted global predictor.

[0061] Specifically, a first prediction result is obtained, where the first prediction result is a first predicted emotion coefficient, which represents the intensity of the target user's current emotional state; the first predicted emotion coefficient is then judged according to the emotion warning threshold, where the emotion warning threshold is used to determine whether the emotional fluctuation has reached a level that requires intervention, and can be set according to the actual scenario. If the first predicted emotion coefficient exceeds the threshold, it indicates that the user may be experiencing significant emotional fluctuations (such as anxiety, excessive stress, etc.); if the first predicted emotion coefficient is greater than the emotion warning threshold, the first predicted emotion coefficient is subtracted from the emotion warning threshold, and the difference between the two is used as the emotion coefficient deviation. This deviation represents the difference between the user's current emotion and their normal state, and is used to assess the degree of current emotional fluctuations. The ratio of the emotion coefficient deviation to the maximum emotion coefficient deviation (which can be set based on historical data) is further calculated, and the ratio is set as the emotional fluctuation scale.

[0062] Then, the scale of the emotional fluctuation is rounded and multiplied by Q to obtain the number of calls to the local prediction branch, that is, the greater the emotional fluctuation, the greater the number of prediction branches used. The initial number of call branches adapted to the local predictor is 1, that is, only one branch participates in the prediction in the initial state of the system to save computing resources. Further, among the Q local prediction branches, the system will randomly select branches with the same number of calls for continuous monitoring. This can ensure the diverse use of multiple branches while avoiding calling all branches every time, saving computing resources. Each time monitoring is performed, the system will record the emotional coefficient deviation and compare it. If the prediction result (emotion coefficient) exceeds the emotional warning threshold, it is considered that an overflow has occurred. During the monitoring process, the number of local prediction branches will be iteratively updated according to the emotional coefficient deviation in each monitoring result.

[0063] During the continuous monitoring process, if the emotion coefficient exceeds the warning threshold multiple times, the number of consecutive overflows is counted, that is, the number of consecutive occurrences of emotional anomalies. When the number of consecutive overflows reaches a predetermined limit (e.g., 3 times), it indicates that the user's emotional anomaly is relatively persistent and cannot be resolved by the local predictor. At this time, the system will activate a more accurate emotion prediction model and activate the adapted global predictor. That is, when the emotion fluctuations of the local predictor continue to exceed the predetermined threshold and after monitoring the number of consecutive overflows, the system will automatically switch to the global predictor for more accurate emotion analysis. Then, according to the global vital sign indicators, the global sign data of the target user is collected and input into the adapted global predictor, and a second prediction result (second predicted emotion coefficient) is output as the emotion monitoring result of the target user.

[0064] The embodiment of the present invention provides an optimization method for emotion prediction based on vital sign data, which has at least the following technical effects:

[0065] By adopting the local emotion prediction model for short-term and efficient monitoring, we can quickly respond to emotional anomalies, and switch to the global emotion prediction model for accurate analysis when the abnormal emotion persists. This can achieve rapid response and in-depth analysis under different emotional states, thereby effectively improving the accuracy and reliability of emotion prediction while ensuring real-time and efficiency, and saving computing power resources.

[0066] For example 2, please refer to Figure 2 , Figure 2 Schematic diagram of an embodiment of an electronic device provided by an embodiment of the present invention. Figure 2 As shown, an embodiment of the present invention provides an electronic device 500, including a memory 510, a processor 520, and a first computer program 511 stored in the memory 510 and executable on the processor 520. When the processor 520 executes the first computer program 511, the following steps are implemented: building an emotion prediction plug-in matching table based on a predetermined basic parameter interval, a predetermined motion parameter interval, and a predetermined environmental parameter interval, wherein the emotion prediction plug-in includes a local emotion predictor and a global emotion predictor, the local emotion predictor being constructed based on a local vital sign indicator, and the global emotion predictor being constructed based on a global vital sign indicator; inputting the target user's basic data, expected motion data, and environmental data into the emotion prediction plug-in matching table to obtain an adapted local predictor and an adapted global predictor; activating the adapted local predictor, collecting local vital sign data according to the local vital sign indicator, inputting the data into the adapted local predictor, and outputting a first prediction result; if the first prediction result exceeds an emotion warning threshold and the number of consecutive overflows reaches a predetermined limit, activating the adapted global predictor, collecting global vital sign data according to the global vital sign indicator, and inputting the data into the adapted global predictor, and outputting a second prediction result as an emotion monitoring result of the target user.

[0067] For example three, please refer to Figure 3 , Figure 3 Schematic diagram of an embodiment of a computer-readable storage medium provided in an embodiment of the present invention. Figure 3 As shown, this embodiment provides a computer-readable storage medium 600, on which a second computer program 611 is stored. When the second computer program 611 is executed by a processor, the following steps are implemented: building an emotion prediction plug-in matching table based on a predetermined basic parameter interval, a predetermined motion parameter interval, and a predetermined environmental parameter interval, wherein the emotion prediction plug-in includes a local emotion predictor and a global emotion predictor, the local emotion predictor being constructed based on a local vital sign indicator, and the global emotion predictor being constructed based on a global vital sign indicator; inputting the target user's basic data, expected motion data, and environmental data into the emotion prediction plug-in matching table to obtain an adapted local predictor and an adapted global predictor; activating the adapted local predictor, collecting local vital sign data according to the local vital sign indicator, inputting the data into the adapted local predictor, and outputting a first prediction result; if the first prediction result exceeds an emotion warning threshold and the number of consecutive overflows reaches a predetermined number limit, activating the adapted global predictor, collecting global vital sign data according to the global vital sign indicator, and inputting the data into the adapted global predictor, and outputting a second prediction result as an emotion monitoring result of the target user.

[0068] It should be noted that, in the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0069] It will be understood by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0070] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded computer, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0071] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0072] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0073] Although preferred embodiments of the present invention have been described, additional changes and modifications to these embodiments may occur to those skilled in the art once the basic inventive concepts become known.

[0074] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the present invention and its equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for optimizing emotion prediction based on vital sign data, characterized in that: Methods include: Building an emotion prediction plug-in matching table based on a predetermined basic parameter interval, a predetermined motion parameter interval, and a predetermined environmental parameter interval, wherein the emotion prediction plug-in includes a local emotion predictor and a global emotion predictor, the local emotion predictor being constructed based on local vital sign indicators, and the global emotion predictor being constructed based on global vital sign indicators; Inputting the target user's basic data, expected motion data, and environmental data into the emotion prediction plug-in matching table to obtain an adapted local predictor and an adapted global predictor; activating the adapted local predictor, collecting local vital sign data according to the local vital sign index, inputting the data into the adapted local predictor, and outputting a first prediction result; If the first prediction result exceeds the emotion warning threshold and the number of consecutive overflows reaches a predetermined limit, activating the adapted global predictor, collecting global vital sign data according to the global vital sign index and inputting it into the adapted global predictor, and outputting a second prediction result as the emotion monitoring result of the target user; Among them, building a global sentiment predictor includes: Configuring a physical sign indicator space, wherein the physical sign indicator space includes a number of physical sign indicators, and the physical sign indicators include at least heart rate, skin galvanic response, respiratory rate, blood oxygen saturation, body temperature, blood pressure, brain waves, etc.; Performing collection convenience and emotion relevance analysis on the individual characteristic indicators, obtaining a number of collection convenience coefficients and a number of emotion relevance coefficients, and performing weighted calculation to obtain a number of indicator influences; Screening the influences of the plurality of indicators according to a predetermined influence threshold, selecting the physical sign indicators corresponding to the influences of the indicators that are greater than the predetermined influence threshold, and obtaining a global physical sign indicator; Based on the principle of ensemble learning, construct the global emotion predictor according to the global vital sign indicators; Wherein, based on the principle of ensemble learning, the global emotion predictor is constructed according to the global vital sign indicators, including: Randomly selecting any parameter from the predetermined basic parameter interval, the predetermined motion parameter interval, and the predetermined environmental parameter interval, and combining them to obtain a first scene parameter; Performing a search based on the first scene parameter and the global vital sign index as constraints to obtain a sample global vital sign index set and a sample emotion coefficient set; Using the sample global vital sign indicator set and the sample emotion coefficient set as training data, supervised training is performed on N emotion prediction operators to obtain N global emotion prediction branches that meet convergence requirements, wherein the emotion prediction operators include at least a BP neural network, a support vector machine, and a random decision forest, and N is an integer greater than or equal to 3; Obtaining N prediction accuracies of the N global emotion prediction branches, and setting N credible weight coefficients according to the N prediction accuracies; Based on the principle of ensemble learning, a first global emotion predictor is constructed according to the N global emotion prediction branches and N credible weight coefficients, and a first mapping association between the first scene parameter and the first global emotion predictor is established.

2. The emotion prediction optimization method based on vital sign data according to claim 1, characterized in that: Build a local sentiment predictor, including: In a two-dimensional space, constructing a correlation fluctuation curve between the global physical sign index and the emotional characteristics to obtain a plurality of global correlation fluctuation curves; performing similarity dimensionality reduction on the multiple global correlation fluctuation curves to obtain multiple difference correlation fluctuation curves, setting the physical sign indicators corresponding to the multiple difference correlation fluctuation curves as difference physical sign indicators, and obtaining multiple difference physical sign indicators; Sort the plurality of difference physical sign indicators according to their influence from large to small, and select the first preset number of indicators in the difference physical sign indicator sequence as local physical sign indicators; Based on the BP neural network, a local emotion predictor is constructed according to the local physical sign indicators.

3. The emotion prediction optimization method based on vital sign data according to claim 2, characterized in that: Performing similarity dimensionality reduction on the multiple global correlation fluctuation curves to obtain multiple difference correlation fluctuation curves, including: randomly selecting a first correlation fluctuation curve from the multiple global correlation fluctuation curves, performing similarity comparisons on the first correlation fluctuation curve and other correlation fluctuation curves from the multiple global correlation fluctuation curves, and obtaining multiple first similarities; Count the number of times the first similarity is less than a predetermined similarity index, and set it as a first difference; If the first difference is greater than a predetermined difference index, the first associated fluctuation curve is set as a difference associated fluctuation curve, and is added to the plurality of difference associated fluctuation curves.

4. The emotion prediction optimization method based on vital sign data according to claim 2, characterized in that: Based on the BP neural network, a local emotion predictor is constructed according to the local physical sign indicators, including: Performing a search based on the first scene parameter and the local physical sign index as constraints to obtain a sample local physical sign index set and a sample emotion coefficient set; Based on the BP neural network, Q local prediction branches are configured, where Q is an integer greater than 1 and less than 5; The sample local sign indicator set and the sample emotion coefficient set are used as a sample data set, and the sample data set is divided into Q equal parts, and Q parts are selected with replacement to construct a first training set, and the selection is iterated Q times to obtain Q training sets; Using the Q training sets, the Q local prediction branches are supervised and trained until convergence, a first local emotion predictor is constructed based on the trained Q local prediction branches, and a second mapping association between the first scene parameter and the first local emotion predictor is established.

5. The method for optimizing emotion prediction based on vital sign data according to claim 4, characterized in that: Based on the first mapping association and the second mapping association, the emotion prediction plug-in matching table is constructed according to the first scene parameter, the first global emotion predictor, and the first local emotion predictor.

6. The emotion prediction optimization method based on vital sign data according to claim 4, characterized in that: If the first prediction result exceeds the emotion warning threshold and the number of consecutive overflows reaches a predetermined limit, activating the adaptive global predictor includes: Obtaining a first prediction result, wherein the first prediction result is a first prediction emotion coefficient; If the first predicted emotion coefficient is greater than the emotion warning threshold, calculating the emotion coefficient deviation, and setting the ratio of the emotion coefficient deviation to the maximum emotion coefficient deviation as the emotion fluctuation scale; The emotional fluctuation scale is rounded and multiplied by Q to obtain the number of calls of the local prediction branch, wherein the initial number of call branches of the adapted local predictor is 1; Randomly select the local prediction branches of the called number from the Q local prediction branches of the adapted local predictor, perform continuous monitoring, and record the number of consecutive overflows, wherein overflow refers to the predicted emotion coefficient being greater than the emotion warning threshold. During the continuous monitoring process, the number of local prediction branches is iteratively updated according to the emotion coefficient deviation in each monitoring result; If the number of consecutive overflows reaches a predetermined limit, the adaptive global predictor is activated.

7. An electronic device, characterized in that: include: Memory for storing computer software programs; A processor is used to read and execute the computer software program, thereby implementing the steps of the emotion prediction optimization method based on vital sign data as described in any one of claims 1 to 6.

8. A non-transitory computer-readable storage medium, characterized in that The storage medium stores a computer software program, which, when executed by a processor, implements the steps of the emotion prediction optimization method based on vital sign data as described in any one of claims 1 to 6.

Images