Psychological assessment method based on emotional fluctuation analysis

Abstract

The invention relates to the technical field of psychological health assessment, in particular to a psychological assessment method based on emotional fluctuation analysis, which comprises the steps of setting a multi-source data acquisition system, performing time synchronization calibration, dynamically adjusting sampling frequency and analysis window length, constructing an emotional fluctuation relation matrix and a nonlinear regression fitting model and the like. According to the method, multi-dimensional data such as the physiological signals, the voice features and the text content are fused, the emotion fluctuation key features are extracted, personalized modeling is carried out, the defects in the aspects of complex emotion fluctuation capture and real-time processing in the prior art are overcome, and technical support is provided for accurate evaluation in the field of mental health. The method can be widely applied to psychological health assessment, multi-modal data fusion sentiment analysis and real-time psychological state monitoring scenes.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of mental health assessment and sentiment analysis, specifically a psychological assessment method based on sentiment fluctuation analysis. BACKGROUND

[0002] The rapid development of psychology and sentiment analysis technology has promoted the importance of psychological assessment methods based on sentiment fluctuation analysis in the field of mental health. By combining sentiment recognition and psychological modeling technology, it can achieve accurate capture of individual emotional state and dynamic assessment of mental health. However, existing technologies still face certain challenges in the depth analysis of sentiment fluctuations, personalized modeling, and real-time processing, and cannot fully meet the actual needs of psychological assessment.

[0003] After searching, a personalized sentiment analysis method and device for sparse data (publication date: April 28, 2023) with publication number CN111325027B focuses on sentiment analysis of text data, and focuses on solving the problem of data sparsity, and cannot fully integrate multi-dimensional sentiment fluctuations (such as physiological signals, speech features, etc.), which may not fully reflect the changes in individual psychological state. In addition, its personalized modeling relies on user grouping strategy, which may lead to insufficient accuracy of the analysis results in the case of large individual differences.

[0004] Another patent with publication number CN109472290B (publication date: October 14, 2022) proposes a sentiment fluctuation model analysis method based on finite state machine. This method collects physiological data and extracts features, combines a support vector machine classifier for emotion classification and recognition, describes the emotion state transition process, and uses psychological questionnaire survey and PAD model for supervised learning training to complete the prediction of emotion fluctuation and emotion quality. However, this technical solution mainly relies on the finite state machine model, and the ability to capture non-linear changes in complex sentiment fluctuations is limited, and the modeling of sentiment state transition is relatively simple, which may affect the accuracy of emotion fluctuation analysis. In addition, this method requires a large amount of psychological questionnaire data as training basis, which may cause data acquisition difficulties in practical application, thereby affecting its generalizability and real-time performance.

[0005] The aforementioned problems indicate that existing techniques based on emotion fluctuation analysis still have room for improvement in areas such as multi-dimensional emotion data integration, complex emotion fluctuation modeling, and real-time processing. Therefore, this invention provides a psychological assessment method based on emotion fluctuation analysis, aiming to construct a more refined emotion fluctuation model by integrating multi-source emotion data (such as physiological signals, speech features, and text content), thereby improving the comprehensiveness and accuracy of psychological assessment, while optimizing real-time analysis capabilities to meet the needs of the mental health field for efficient and accurate psychological assessment. Summary of the Invention

[0006] To address the aforementioned issues, this invention aims to propose a psychological assessment method and application based on emotional fluctuation analysis. By introducing an emotional fluctuation modeling mechanism that integrates multi-source data, a dynamic change model of an individual's emotional state is constructed. By combining multi-dimensional data such as physiological signals, speech features, and text content, key features of emotional fluctuations are extracted and nonlinear modeled, thereby overcoming the shortcomings of existing technologies in capturing complex emotional fluctuations, personalized modeling, and real-time processing, and providing technical support for accurate assessment in the field of mental health.

[0007] To achieve the above objectives, the technical solution of the present invention is implemented as follows:

[0008] A psychological assessment method based on emotion fluctuation analysis includes the following steps: setting up a multi-source data acquisition system comprising a physiological signal acquisition module, a speech feature extraction module, and a text content analysis module; performing time synchronization calibration between the physiological signal acquisition module and the speech feature extraction module to establish a unified time reference link; dynamically adjusting the sampling frequency of the physiological signal acquisition module and the analysis window length of the speech feature extraction module to fully adapt to the dynamic characteristics of different emotion fluctuations; setting a movable emotion simulation unit at the center of the multi-source data acquisition system, adjusting its relative distance to each acquisition module, and acquiring multiple sets of emotion fluctuation samples using step-by-step measurement; constructing an emotion fluctuation relationship matrix comprising two variables, sampling frequency and relative distance, through repeated measurements under multiple sampling frequency conditions; and using a nonlinear regression algorithm to perform bivariate joint fitting on the measurement data to establish an emotion fluctuation model with personalized compensation.

[0009] Furthermore, a physiological signal acquisition module, a speech feature extraction module, and a text content analysis module are set up. The sampling frequency between the physiological signal acquisition module and the speech feature extraction module is adjustable, and the text content analysis module is placed on the logical connection path between the physiological signal acquisition module and the speech feature extraction module.

[0010] Furthermore, the physiological signal acquisition module and the speech feature extraction module are connected to the central processing unit through a high-speed data transmission interface, and both the physiological signal acquisition module and the speech feature extraction module are equipped with timestamp marking function, thereby establishing a time reference link.

[0011] Further, the sampling frequency of the physiological signal acquisition module and the analysis window length of the speech feature extraction module are adjusted to adapt to the dynamic characteristics of different emotional fluctuations.

[0012] Further, the sampling frequency between the physiological signal acquisition module and the speech feature extraction module is fixed, the relative distance between the emotion simulation unit and each acquisition module is adjusted, the emotional fluctuation data of the emotion simulation unit at the relative distance is collected by the emotional fluctuation measuring instrument, and the average value of the emotional fluctuation intensity is calculated from the data as the emotional fluctuation intensity value corresponding to each measurement.

[0013] Further, the sampling frequency between the physiological signal acquisition module and the speech feature extraction module is changed and the value of the relative distance is adjusted, the above steps are repeated, and the demarcation value of the relative distance is obtained in the range of 0.1-1.0 meters; a plurality of emotional fluctuation intensity values are obtained according to the above steps, and an emotional fluctuation model with the influence of the relative distance is constructed.

[0014] Further, the emotional fluctuation model constructed above is parameter fitted by a nonlinear regression algorithm to obtain the correction value of each parameter of the emotional fluctuation model, and the final emotional fluctuation model is obtained by using the correction value of each parameter.

[0015] Further, in the step, the obtained emotional fluctuation intensity value is parameter fitted, and the emotional fluctuation index can be expressed as:

[0016] Wherein, d represents the relative distance between the emotion simulation unit and each acquisition module, and a, b, and c respectively represent the coefficients of the above expression;

[0017] And the final expression of the constructed emotional fluctuation model with the influence of the relative distance is:

[0018] Wherein, d represents the relative distance between the emotion simulation unit and each acquisition module; a, b, and c represent the coefficients of the above expression.

[0019] Further, the expression of the final emotional fluctuation model is expressed as:

[0020] Wherein, d0 represents the selected reference distance, represents the emotional fluctuation intensity value at the reference distance d0, represents the sampling frequency between the physiological signal acquisition module and the speech feature extraction module, represents the relative distance between the emotion simulation unit and each acquisition module, represents the emotional fluctuation index, and represents the random noise obeying the normal distribution with the mean value of 0 and the standard deviation of 1.

[0021] In order to achieve the above-mentioned purpose, the application further provides an application of a psychological evaluation method based on emotional fluctuation analysis, which can be used in a psychological health evaluation scene, or used as multi-modal data fusion emotional analysis, or used as real-time psychological state monitoring.

[0022] Beneficial effects: the application introduces a multi-source data fusion emotional fluctuation modeling mechanism to construct a dynamic change model of individual emotional state; by combining physiological signals, speech features and text content and other multi-dimensional data, key features of emotional fluctuation are extracted and nonlinear modeling is performed, so that the deficiencies of the prior art in complex emotional fluctuation capture, personalized modeling and real-time processing are solved, and technical support is provided for precise evaluation in the field of mental health. In addition, the application can be used as multi-modal data fusion emotional analysis (such as a hybrid analysis mechanism combining physiological signals and speech features), and can also be used as real-time psychological state monitoring.

[0023] Figure 1 It is a structural schematic diagram of a multi-source data acquisition system in an embodiment of the application, which shows the logical connection relationship of a physiological signal acquisition module, a speech feature extraction module and a text content analysis module, and the data transmission path of each module and a central processor.

[0024] Figure 2 It is a working scene schematic diagram of an emotional fluctuation measuring instrument in an embodiment of the application, which shows the relative distance adjustment mode between an emotion simulation unit and a physiological signal acquisition module and a speech feature extraction module, and the acquisition process of emotional fluctuation samples in a step-by-step measurement process.

[0025] Figure 3 It is a construction flowchart of an emotional fluctuation model in an embodiment of the application, which describes the process of jointly fitting the sampling frequency and the relative distance double variables by a nonlinear regression algorithm, and shows the expression derivation steps of the final emotional fluctuation model.

[0026] The reference signs are as follows: 1, physiological signal acquisition module; 2, speech feature extraction module; 3, text content analysis module; 4, central processor; 5, emotion simulation unit; 6, time reference link; 7, emotional fluctuation measuring instrument; 8, nonlinear regression algorithm module. SPECIFIC EMBODIMENTS

[0027] The application provides a psychological evaluation method based on emotional fluctuation analysis, which is described below in combination with the accompanying drawings. Figure 1 to the accompanying drawings Figure 3 The specific embodiments of the application are described in detail. Figure 1Fig. 1 is a structural diagram of a multi-source data acquisition system, showing the logical connection relationship between a physiological signal acquisition module 1, a speech feature extraction module 2, and a text content analysis module 3, as well as the data transmission path between each module and a central processor 4. Figure 2 Fig. 2 is a working scenario diagram of an emotional fluctuation measuring instrument 7, showing the relative distance adjustment mode between an emotion simulation unit 5 and the physiological signal acquisition module 1 and the speech feature extraction module 2, as well as the acquisition process of emotional fluctuation samples in a step-by-step measurement process. Figure 3 Fig. 3 is a flowchart of the construction of an emotional fluctuation model, describing the joint fitting process of the sampling frequency and the relative distance double variables through a nonlinear regression algorithm 8, and showing the expression derivation steps of the final emotional fluctuation model.

[0028] In actual application, first, a multi-source data acquisition system containing a physiological signal acquisition module 1, a speech feature extraction module 2, and a text content analysis module 3 is set. The physiological signal acquisition module 1 is used to acquire physiological signals of an individual, such as heart rate, galvanic skin response, and brain waves, etc.; the speech feature extraction module 2 is used to extract features such as pitch, speech rate, and energy distribution from the speech of the individual; and the text content analysis module 3 is responsible for emotional analysis of the text content input by the individual, extracting information such as keywords and emotional tendency, etc. The above three modules are connected to the central processor 4 through a high-speed data transmission interface, and the central processor 4 is responsible for integrating data from each module and subsequent processing. In order to ensure time synchronization, the physiological signal acquisition module 1 and the speech feature extraction module 2 are both equipped with timestamp marking functions, and a unified time reference is established through a time reference link 6, thereby ensuring the time consistency of multi-source data.

[0029] In Figure 1 , the sampling frequency between the physiological signal acquisition module 1 and the speech feature extraction module 2 is adjustable, and the text content analysis module 3 is placed on the logical connection path between the physiological signal acquisition module 1 and the speech feature extraction module 2. This design enables the system to flexibly adjust the sampling frequency and the analysis window length according to the dynamic characteristics of emotional fluctuation. For example, when capturing rapidly changing emotional fluctuations, the sampling frequency can be increased to obtain more detailed data; and when analyzing slower changing emotional states, the sampling frequency can be reduced to reduce data redundancy. In addition, the introduction of the text content analysis module 3 enables the system to comprehensively consider the emotional information in language expression, thereby improving the accuracy of emotional fluctuation modeling.

[0030] Next, the emotion simulation unit 5 is set at the center position of the multi-source data acquisition system, and the relative distance between it and the physiological signal acquisition module 1 and the speech feature extraction module 2 is adjusted. The role of the emotion simulation unit 5 is to simulate the behavior of individuals in different emotional states, providing standardized emotional fluctuation samples for the system. In Figure 2In the embodiment, the relative distance between the emotion simulation unit 5 and each acquisition module is adjusted by step measurement, and a set of emotional fluctuation data is collected by using the emotional fluctuation measuring instrument 7 after each adjustment. In the specific operation, the sampling frequency between the physiological signal acquisition module 1 and the speech feature extraction module 2 is first fixed, and then the relative distance between the emotion simulation unit 5 and each acquisition module is gradually changed, and the range is set to 0.1 meters to 1.0 meters. In this process, the emotional fluctuation intensity value corresponding to each measurement is recorded, and the average value is calculated as the emotional fluctuation intensity at the relative distance.

[0031] After completing the step measurement, the sampling frequency between the physiological signal acquisition module 1 and the speech feature extraction module 2 is changed and the above steps are repeated to obtain emotional fluctuation data under different sampling frequencies. Through multiple measurements, an emotional fluctuation relationship matrix containing two variables of sampling frequency and relative distance is constructed. In Figure 3 In the embodiment, the construction process of the emotional fluctuation relationship matrix is fitted by a nonlinear regression algorithm 8 with two variables, and finally an emotional fluctuation model with personalized compensation is obtained. Specifically, the emotional fluctuation index can be expressed as a function, where d represents the relative distance between the emotion simulation unit 5 and each acquisition module, and a, b, and c represent the coefficients of the above expression. Further, the final expression of the constructed emotional fluctuation model with the influence of the relative distance is: E (t) = a + b f + c d + e, where d0 is the selected reference distance, E (d0) represents the emotional fluctuation intensity value at the reference distance, f represents the sampling frequency between the physiological signal acquisition module 1 and the speech feature extraction module 2, d represents the relative distance between the emotion simulation unit 5 and each acquisition module, E (t) represents the emotional fluctuation index, and e represents random noise obeying normal distribution with mean value of 0 and standard deviation of 1.

[0032] In actual operation, the central processor 4 receives multi-source data from the physiological signal acquisition module 1, the speech feature extraction module 2 and the text content analysis module 3, and performs time synchronization calibration through the time reference link 6. Subsequently, the central processor 4 inputs the synchronized data into the nonlinear regression algorithm module 8, uses the preset emotional fluctuation model to fit the data, and obtains the final emotional fluctuation intensity value. This process not only can capture complex emotional fluctuations, but also can conduct personalized modeling according to individual differences, thereby improving the accuracy of psychological evaluation.

[0033] The application scenarios of the present application include mental health assessment, multi-modal data fusion emotional analysis, and real-time psychological state monitoring. In the mental health assessment scenario, the system can generate a dynamic emotional fluctuation curve by analyzing the individual's physiological signals, speech features, and text content, helping the psychologist better understand the individual's emotional state. In the multi-modal data fusion emotional analysis, the system can combine physiological signals and speech features for mixed analysis, thereby improving the accuracy of emotional recognition. In real-time psychological state monitoring, the system can continuously collect multi-source data of individuals and update the emotional fluctuation model in real time to provide timely psychological health warnings for individuals.

[0034] As can be seen from the above embodiments, the present application solves the deficiencies of the prior art in complex emotional fluctuation capture, personalized modeling, and real-time processing by using a multi-source data fusion emotional fluctuation modeling mechanism that combines physiological signals, speech features, and multi-dimensional data such as text content, thereby providing technical support for precise assessment in the field of mental health. In order to better enable relevant personnel in the technical field to fully understand and implement the present application, the specific implementation principles of the present application are further supplemented below in conjunction with a specific application scenario.

[0035] In the actual scenario of mental health assessment, first, the physiological signal acquisition module 1 acquires the individual's heart rate, skin electrical response, and brain wave physiological data in real time. These data have high time sensitivity, so in the acquisition process, a time stamp is marked for each set of data through the time reference link 6 to ensure synchronization with the data of other modules. At the same time, the speech feature extraction module 2 extracts features such as pitch, speech rate, and energy distribution from the individual's speech input and transmits the data to the central processor 4 through the high-speed data transmission interface. The text content analysis module 3 performs emotional analysis on the received text content, extracts keywords and emotional tendency information, and integrates it with the physiological signal and speech feature data. The above-mentioned multi-source data acquisition and transmission process is completed through the logical connection relationship shown in Figure 1 The sampling frequency can be flexibly adjusted according to the dynamic characteristics of emotional fluctuation to ensure that the fineness and redundancy of the data reach the best balance.

[0036] In actual operation, the emotion simulation unit 5 is placed in the central position of the multi-source data acquisition system, and the Figure 2The relative distance between the physiological signal acquisition module 1 and the speech feature extraction module 2 is adjusted in the manner shown. Specifically, first, the sampling frequency between the physiological signal acquisition module 1 and the speech feature extraction module 2 is fixed, and then the relative distance between the emotion simulation unit 5 and each acquisition module is gradually changed, ranging from 0.1 meters to 1.0 meters. After each adjustment, a set of emotional fluctuation data is collected using the emotional fluctuation measuring instrument 7, and the emotional fluctuation intensity value corresponding to each measurement is recorded. Through multiple measurements, the average emotional fluctuation intensity value under different relative distances is calculated, thereby constructing a preliminary emotional fluctuation sample set. Subsequently, the sampling frequency between the physiological signal acquisition module 1 and the speech feature extraction module 2 is changed and the above steps are repeated to obtain emotional fluctuation data under different sampling frequencies. This process is completed through step-by-step measurement to ensure the comprehensiveness and accuracy of the data.

[0037] After obtaining sufficient emotional fluctuation samples, a nonlinear regression algorithm 8 is used to jointly fit the sampling frequency and relative distance double variables to construct an emotional fluctuation model. As shown in Figure 3 The model describes the relationship between the emotional fluctuation index and the sampling frequency and relative distance through a mathematical expression. Specifically, the emotional fluctuation index can be expressed as: E = A ⋅ f2 + B ⋅ d + C E = A ⋅ f2 + B ⋅ d + C where f represents the sampling frequency, d represents the relative distance between the emotion simulation unit 5 and each acquisition module, A, B, and C are fitting coefficients. The final emotional fluctuation model expression is: E = E0 ⋅ exp(−σd−d0) + ϵ E = E0 ⋅ exp(−σd−d0) + ϵ where E0 represents the emotional fluctuation intensity value at the reference distance d0, σ is the attenuation coefficient, and ϵ is a normally distributed random noise with mean 0 and standard deviation σϵ. This model can compensate for individual differences, thereby improving the accuracy of emotional fluctuation analysis.

[0038] In actual operation, the central processor 4 receives multi-source data from the physiological signal acquisition module 1, the speech feature extraction module 2, and the text content analysis module 3, and performs time synchronization calibration through the time reference link 6. Subsequently, the central processor 4 inputs the synchronized data into the nonlinear regression algorithm module 8, uses the pre-set emotional fluctuation model to fit the data, and obtains the final emotional fluctuation intensity value. This process not only captures complex emotional fluctuations, but also builds personalized models according to individual differences, thereby improving the accuracy of psychological assessment.

[0039] In the application scenario of mental health assessment, the system analyzes the individual's physiological signals, speech features and text content to generate a dynamic emotional fluctuation curve, helping psychologists better understand the individual's emotional state changes. For example, when the individual shows a faster heart rate, a higher voice pitch and negative emotional words in the text content, the system will automatically identify possible emotional abnormalities and generate corresponding warning information. In multi-modal data fusion emotional analysis, the system combines physiological signals and speech features for mixed analysis to improve the accuracy of emotional recognition. For example, by comparing the individual's physiological signal and speech feature changes at different times, the system can more accurately judge the emotional fluctuation trend. In real-time mental state monitoring, the system can continuously collect multi-source data of the individual and update the emotional fluctuation model in real time to provide timely psychological health warnings for the individual. For example, when the system detects that the individual's emotional fluctuation intensity exceeds the preset threshold, it will automatically trigger the alarm mechanism and send a notification to the relevant personnel.

[0040] As can be seen from the above specific application scenarios, the present application solves the deficiencies of the prior art in complex emotional fluctuation capture, personalized modeling and real-time processing by using a multi-source data fusion emotional fluctuation modeling mechanism, combining multi-dimensional data such as physiological signals, speech features and text content, and providing technical support for precise assessment in the field of mental health.

Claims

1. A psychological assessment method based on emotion fluctuation analysis, characterized in that, Includes the following steps: S1: Set up a multi-source data acquisition system including a physiological signal acquisition module (1), a speech feature extraction module (2), and a text content analysis module (3); S2: Perform time synchronization calibration between the physiological signal acquisition module (1) and the speech feature extraction module (2) to establish a unified time reference link (6); S3: Dynamically adjust the sampling frequency of the physiological signal acquisition module (1) and the analysis window length of the speech feature extraction module (2) to adapt to the dynamic characteristics of different emotional fluctuations; S4: Set a movable emotion simulation unit (5) at the center of the multi-source data acquisition system, adjust its relative distance with each acquisition module, and use step-by-step measurement to obtain multiple sets of emotional fluctuation samples; Step S4 specifically refers to: fixing the sampling frequency between the physiological signal acquisition module (1) and the speech feature extraction module (2), adjusting the emotional fluctuations, and adjusting the sampling frequency between the physiological signal acquisition module (1) and the speech feature extraction module (2) to adapt to the dynamic characteristics of different emotional fluctuations. The relative distance between the simulation unit (5) and each acquisition module is used to collect emotional fluctuation data of the emotional simulation unit (5) at its relative distance using the emotional fluctuation measurement instrument (7), and the average value of the emotional fluctuation intensity is calculated from this data as the emotional fluctuation intensity value corresponding to each measurement; S5: Through repeated measurements under multiple sampling frequency conditions, an emotional fluctuation relationship matrix containing the bivariate of sampling frequency and relative distance is constructed; The step S5 specifically refers to: changing the sampling frequency between the physiological signal acquisition module (1) and the speech feature extraction module (2) and adjusting the value of the relative distance, repeating the above step S4, and obtaining the relative distance in the range of 0.1 meters to 1.0 meters. Boundary value; Based on the above steps, multiple emotional fluctuation intensity values ​​are obtained, and an emotional fluctuation model with relative distance influence is constructed; S6: A nonlinear regression algorithm is used to perform bivariate joint fitting on the measurement data to establish an emotional fluctuation model with personalized compensation; Specifically, step S6 refers to: using a nonlinear regression algorithm to fit the parameters of the emotional fluctuation model constructed in step S5 above, obtaining the correction values ​​of each parameter of the emotional fluctuation model, and using the correction values ​​of each parameter to obtain the final emotional fluctuation model; In step S6, the obtained emotional fluctuation intensity values ​​are fitted with parameters, and the emotional fluctuation index can be expressed as: where, represents the distance between the emotion simulation unit (5) and each acquisition module. The relative distance, and, respectively represent the coefficients of the above expressions; and the final expression of the constructed emotional fluctuation model with relative distance influence is: where, is the relative distance between the emotion simulation unit (5) and each acquisition module; is the coefficient of the above expressions, and the expression of the final emotional fluctuation model is: where, is the selected reference distance, represents the emotional fluctuation intensity value at the reference distance, represents the sampling frequency between the physiological signal acquisition module (1) and the speech feature extraction module (2), represents the relative distance between the emotion simulation unit (5) and each acquisition module, represents the emotional fluctuation index, and is random noise that follows a normal distribution with a mean of 0 and a standard deviation of.

2. The psychological assessment method based on emotional fluctuation analysis according to claim 1, characterized in that, The step S1 specifically refers to setting up a physiological signal acquisition module (1), a speech feature extraction module (2), and a text content analysis module (3), wherein the sampling frequency between the physiological signal acquisition module (1) and the speech feature extraction module (2) is adjustable, and the text content analysis module (3) is placed on the logical connection path between the physiological signal acquisition module (1) and the speech feature extraction module (2).

3. The psychological assessment method based on emotional fluctuation analysis according to claim 1, characterized in that, The step S2 specifically refers to the following: the physiological signal acquisition module (1) and the speech feature extraction module (2) are connected to the central processing unit (4) through a high-speed data transmission interface, and both the physiological signal acquisition module (1) and the speech feature extraction module (2) are equipped with timestamp marking function, thereby establishing a time reference link (6).

Images