A virtual reality exposure therapy method and system based on multi-modal data feedback

Abstract

The application provides a virtual reality exposure therapy method and system based on multi-modal data feedback, aiming to optimize anxiety treatment through dynamic and closed-loop control. The core steps are as follows: first, in the VR exposure, the multi-modal data such as physiological signals, behavior interaction and subjective evaluation of the user are synchronously collected. Then, based on these data, the individualized emotional portrait representing the emotional state of the user is constructed and updated in real time. Further, the quantitative emotional load index EmoLoad is generated according to the portrait to evaluate the real-time emotional load state of the user. Finally, the system automatically and dynamically adjusts the difficulty of the VR exposure scene by comparing the EmoLoad with the preset treatment target interval, so as to guide the emotional load of the user to the ideal treatment range, forming a personalized closed-loop intervention. The application integrates immersive environment, biological feedback and intelligent analysis to break through the scene and personalization limitations of traditional exposure therapy.

Description

Technical Field

[0001] This invention relates to the interdisciplinary field of virtual reality technology and clinical psychology and psychiatry, specifically to a virtual reality exposure therapy method and system based on multimodal data feedback. Background Technology

[0002] Traditional exposure therapy requires patients to gradually encounter anxiety-provoking situations in real-world environments, such as classroom participation, peer interaction, or entering a school. However, key triggering situations (such as school classrooms, exam scenarios, and classmate gatherings) are difficult to directly recreate in the treatment setting. Patients need to be exposed on campus or in specific scenarios, which is often limited by time, space, venue, and third-party cooperation, making treatment implementation difficult. Furthermore, these real-world scenarios are often difficult to fully control in practice (e.g., peer reactions, school environment factors), making it difficult to guarantee a gradual and controllable exposure process. Many patients strongly resist real-world exposure, often exhibiting avoidance behaviors, leading to treatment interruptions or limited effectiveness. Organizing and implementing exposure training in real-world environments often requires additional human, material, and time costs (e.g., taking patients to school, simulating classroom participation scenarios, etc.). Therapists, with limited clinical time, find it difficult to repeatedly perform high-frequency exposures. Finally, these shortcomings limit the effectiveness of practical clinical application and restrict its large-scale promotion and standardized implementation.

[0003] Virtual reality and other extended reality technologies have been widely used in entertainment, education and other fields, but they still have limitations in clinical applications in mental and psychological therapy, and are still insufficient in terms of personalized adaptation and single assessment dimensions.

[0004] 1) Current VR therapy systems mostly employ fixed scene designs, pre-setting typical scenarios but failing to dynamically adjust based on the patient's real-time state and individual differences. For example, different patients have different anxiety and fear sensitivities, and their emotional state changes during treatment vary, resulting in a low degree of matching between the established treatment plan and the individual patient's needs. Furthermore, due to regional cultural differences, there is insufficient data on the recording, statistics, and research of differences in emotional exposure feedback, making it unsuitable for rigorous sample studies.

[0005] 2) Existing virtual reality technology applied to exposure therapy assessment systems lacks the collection, analysis, and presentation of diverse, multimodal data, failing to provide healthcare professionals with real-time monitoring and understanding of patient exposure. Post-treatment assessments also overly rely on patient subjective evaluations, lacking real-time feedback and integration of physiological indicators and behavioral data, and failing to provide data and information support for evaluating or studying changes in patients during treatment. Furthermore, there is a lack of dynamic adjustment mechanisms. Unlike traditional exposure therapy, which is guided throughout by healthcare professionals and adjusted based on whether there is excessive exposure or insufficient stimulation, current digital applications cannot replace or assist healthcare professionals in making judgments. They primarily rely on therapist observation and client self-reports to assess efficacy, resulting in strong subjectivity, insufficient real-time data, and a lack of objective quantification, thus limiting the accuracy and reliability of efficacy assessments. Summary of the Invention

[0006] The purpose of this invention is to provide a virtual reality exposure therapy method and system based on multimodal data feedback, which integrates strategies such as immersive virtual environment construction, biofeedback mechanisms and cognitive behavioral analysis to achieve systematic clinical intervention for anxiety.

[0007] To achieve the above objectives, the present invention provides a virtual reality exposure therapy method based on multimodal data feedback, which presents an exposure scenario and collects user data through a virtual reality device. The method includes the following closed-loop steps: S1, Synchronous Acquisition and Processing of Multimodal Data: During virtual reality exposure, the system simultaneously collects users' physiological signal data and behavioral interaction data, and receives subjective evaluation data input by users. S2, Real-time Construction and Update of Personalized Emotional Profiles: Based on the physiological signal data, behavioral interaction data, and subjective evaluation data, an emotional feature model is constructed to characterize the user's emotional state. S3, Emotional Load Assessment Based on the emotional feature model, an emotional load index, EmoLoad, is generated to characterize the user's emotional load state during virtual reality exposure. S4, Decision Output and Dynamic Difficulty Adjustment Based on the determination result of the emotional load index EmoLoad relative to the preset emotional load range, the difficulty level of virtual reality exposure therapy is automatically adjusted so that the user's real-time emotional load is guided to the treatment target range, thereby forming an individualized and dynamic closed-loop exposure control.

[0008] Furthermore, the physiological signal data, behavioral interaction data, and subjective evaluation data are timestamped based on a unified system time reference, and after time alignment, they are segmented into segments with a fixed-length sliding time window, wherein the duration of the sliding time window is 3 seconds. Furthermore, within each of the sliding time windows, the physiological signal data is processed to calculate the following physiological characteristic indicators: ΔRMSSD is used to represent the magnitude of change in the heart rate variability index relative to the user's physiological baseline within the current time window; ΔHF is used to represent the magnitude of change of the high-frequency component of heart rate variability relative to the physiological baseline within the current time window; SCL slope is used to represent the slope of the change of the slow-varying component of the skin conductance signal within the time window. The Heart Rate Response Index (HRI) is used to indicate the degree of deviation of the current heart rate level from the physiological baseline.

[0009] Furthermore, the ΔRMSSD, ΔHF, SCL slope, and HRI are normalized and comprehensively calculated according to a preset weight relationship to construct the BioArousal index, which characterizes the arousal level of the user's autonomic nervous system.

[0010] Furthermore, based on behavioral interaction data in the virtual reality environment, the following behavioral features are extracted: The gaze interruption ratio represents the percentage of time a user's gaze deviates from the target stimulus within a preset time window. Backward motion speed, which represents the user's displacement speed in the virtual reality environment along the direction away from the stimulus; The number of rapid head shifts indicates the number of times the user's head posture changes beyond a preset angular velocity threshold per unit time. The behavioral characteristics are then normalized and weighted to construct the AvoidScore, a behavioral avoidance index.

[0011] Furthermore, the subjective assessment data consists of the subjective pain or emotional intensity ratings input by the user during virtual reality exposure. These ratings are divided into multiple level ranges according to a preset scale and converted into corresponding numerical features for the calculation of emotional load indicators.

[0012] Furthermore, the generation of the emotional load index EmoLoad includes the following steps: The numerical characteristics of the BioArousal index, AvoidScore index, and subjective evaluation data are normalized to map them to a uniform numerical range. Based on preset weight parameters, the normalized features are weighted and calculated to obtain their respective weighted feature values. Based on the weighted feature values, the weighted features are comprehensively calculated according to the preset fusion rules to generate the emotional load index EmoLoad; The emotional load index EmoLoad is compared with a preset emotional load threshold range to determine whether the user's current emotional state is in the tolerable zone, the edge overload zone, the overload zone, or the low load zone.

[0013] Furthermore, when the EmoLoad indicator is detected to enter the preset edge zone or overload zone, the system automatically performs real-time difficulty adjustment operations, including reducing the number of stimuli, blurring visual stimuli, increasing safety cues, briefly pausing the enhanced context, and delaying the division of the next contextual event, thereby forming an individualized and dynamic closed-loop exposure control process.

[0014] Furthermore, when the EmoLoad indicator is detected to be continuously in the low load zone, the system automatically performs real-time difficulty adjustment operations, including reducing the number of stimuli, blurring visual stimuli, increasing the distance of stimuli, and increasing scene immersion or interaction complexity, thereby forming an individualized and dynamic closed-loop exposure control process.

[0015] On the other hand, the present invention also provides a virtual reality exposure therapy system based on multimodal data feedback, comprising: The virtual reality interaction module is used to present exposed scenes and collect user behavior and interaction data; A multimodal data acquisition module is used to collect users' physiological signal data; The subjective evaluation input module is used to receive subjective evaluation data from users during the exposure process; The data synchronization and processing module is used to perform time alignment and sliding window processing on the physiological signal data, behavioral interaction data and subjective evaluation data; The emotion modeling module is used to build individualized emotion profiles and generate the emotion load index EmoLoad; The decision-making and control module is used to automatically adjust the difficulty level of virtual reality exposure scenarios based on the EmoLoad index.

[0016] The present invention has the following beneficial effects and technological advancements. 1. Existing VR exposure methods are mostly concentrated in single public speaking or classroom scenarios, employing fixed scene designs and pre-set typical situations, but failing to dynamically adjust according to the patient's real-time state and individual patient differences. This invention achieves personalized treatment through dynamic scene adjustment. By observing the patient's multimodal physiological and behavioral data, personalized treatment is achieved through dynamic scene adjustment, greatly reducing the shortcomings of traditional VR exposure and effectively improving treatment effectiveness.

[0017] 2. Existing virtual reality technology applied to exposure therapy assessment systems lacks the collection, analysis, and presentation of diverse, multimodal data, failing to provide healthcare professionals with real-time monitoring and understanding of patient exposure. Furthermore, post-treatment assessments rely excessively on patient subjective evaluations, lacking real-time feedback and integration of physiological indicators and behavioral data. This technology aims to collect relevant data through a multimodal data acquisition module, combine it with self-evaluation results, and conduct data integration and analysis. Through research, it seeks to obtain effective physiological and behavioral data indicators for exposure, thereby achieving standardized and repeatable physiological and behavioral data indicators for effective exposure. This eliminates reliance on patient subjective reports, significantly improving the effectiveness of exposure implementation. It not only provides objective, real-time, and traceable quantitative evidence for exposure efficacy but also enables individualized intervention and standardized promotion, thereby improving efficacy, reducing costs, and potentially establishing evidence-based standards in this field. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of a virtual reality exposure therapy method based on multimodal data feedback according to the present invention.

[0020] Figure 2 This is a system framework diagram of a virtual reality exposure therapy system based on multimodal data feedback according to the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] To better understand the above technical solutions, exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present invention and to fully convey the scope of the invention to those skilled in the art.

[0023] The method and system proposed in this invention aim to combine and realize the advantages of traditional manual exposure therapy, further enhancing its applicability and efficacy based on current applications of virtual reality. Traditional exposure therapy uses both real-world and imagined exposure methods. Compared to existing digital exposure therapies, it still has the advantages of personalized exposure scenarios and the ability of medical staff to monitor and intervene in patients' emotions at any time. This invention addresses and improves upon the shortcomings of existing technologies in handling the flexibility of personalized design and the lack of patient emotional and behavioral data collection.

[0024] Figure 1 This is a flowchart of a virtual reality exposure therapy method based on multimodal data feedback according to the present invention. Figure 1 The present invention presents an exposure scenario and collects user data through a virtual reality device, forming a closed-loop process of data integration and decision-making.

[0025] In this embodiment, the hardware system adopts a collaborative architecture of "wearable VR device + physiological data acquisition device", and the specific configuration is as follows: Virtual Reality Interaction Equipment: Immersive wearable VR devices are selected, featuring high-definition visual presentation, head motion capture, and controller interaction control functions. These devices are used to construct exposure scenarios (such as classroom speaking, social gatherings, and specific fear-inducing scenarios) and collect user behavior interaction data in the virtual environment (viewpoint rotation, head movement speed, controller operation frequency, spatial position changes, etc.).

[0026] Physiological data acquisition equipment: Equipped with a professional physiological signal acquisition module, including an electrocardiogram (ECG) acquisition sensor and an electrical conductance oscillation (EDA) acquisition sensor, which are used to acquire the user's ECG and EDA signals respectively. The sampling rate is set to 1000Hz to meet the time resolution requirements and realize dynamic monitoring of physiological responses during fear extinction.

[0027] The data synchronization and processing unit is equipped with a high-performance processor and data transmission module, which supports the synchronous reception, timestamp marking, preprocessing and feature extraction of multimodal data, ensuring the efficient integration and real-time analysis of physiological signal data, behavioral interaction data and subjective evaluation data.

[0028] In this embodiment, by fusing and extracting features from electrophysiological signals, behavioral patterns, and subjective self-report data under a unified time reference, an individualized emotional feature model that can be used for dynamic exposure regulation is formed. The specific steps are as follows: S1, synchronous acquisition and processing of multimodal data.

[0029] Physiological signal acquisition and preprocessing: Electrocardiogram (ECG) and electrical conductance analysis (EDA) signals are acquired simultaneously at a sampling rate of 1000Hz. Artifact removal (such as electromyographic interference and motion artifacts), interpolation correction, filtering and smoothing are performed on the raw signals to obtain structured and standardized physiological sequence data.

[0030] ECG signal processing involved identifying and correcting ectopic beats and missing heartbeats using Matlab's HRVtool. Outlier identification, interpolation, filtering, and smoothing were performed on the RR interval (RRI) data, removing data with an outlier ratio greater than 30%. The root mean square (RMSSD) of the RR interval difference, a temporal index of heart rate variability, and the high-frequency component (HF, frequency range 0.15-0.4Hz), a frequency domain index, were calculated. Both are sensitive indicators of parasympathetic nervous system regulation.

[0031] Skin conductance signal processing was performed using Matlab Ledalab. After filtering with a Butterworth low-pass filter with a cutoff frequency of 5Hz and a filter order of 10, the signal was smoothed using an adaptive Gaussian window with a window width of no more than 3 seconds. The slowly varying components were then estimated through continuous decomposition analysis, and the average value of the slowly varying components within a specific time window was used as the skin conductance level (SCL) as a characteristic parameter of skin conductance.

[0032] The collection and preprocessing of behavioral interaction data involves VR devices collecting real-time behavioral data from users in the virtual environment, such as head movement trajectory, gaze point, behavioral reaction time, viewpoint rotation, head movement speed, controller operation frequency, micro-evasion actions, and changes in spatial position. The data is cleaned, event segments are labeled, and time sequence is aligned. Core behavioral features such as "gaze interruption ratio," "backward movement speed," and "number of rapid head shifts" are extracted to construct a feature set that reflects the intensity of behavior and avoidance response during exposure.

[0033] Subjective assessment data collection: During virtual reality exposure, users can input subjective distress units (SUDs) ratings (0-10 points) at any time via VR controllers. The system divides the ratings into multiple level ranges according to a preset scale and converts them into corresponding numerical features for the calculation of emotional load indicators.

[0034] S2, real-time construction and updating of individualized emotional profiles.

[0035] In this embodiment, the method integrates processed physiological features (RMSSD, HF, SCL), behavioral features (gaze interruption ratio, backward movement speed, etc.) with subjective emotion scores at the feature level according to a unified timestamp. The position of these features in the multi-dimensional coordinate space of "arousal degree - avoidance degree - emotional load" is calculated through a weighted feature mapping model to form a dynamically updated emotion representation vector.

[0036] Based on the above emotional representation sequence, an individualized emotional profile of the patient is constructed, including baseline emotional response threshold, stress recovery curve, key triggering factors (such as specific scene stimuli, interaction intensity, etc.) and individualized emotional sensitivity coefficient, providing accurate quantitative basis for real-time adjustment and early warning monitoring of subsequent VR exposure intensity.

[0037] In this embodiment, the time alignment process adds a unified system timestamp to all data, with a 1000Hz time base as the main reference. Different modal signals are resampled to a unified time axis using nearest neighbor interpolation to ensure accurate matching of physiological signals, behavioral signals, and subjective evaluation data in the time dimension.

[0038] After time alignment, physiological signal data, behavioral interaction data, and subjective evaluation data are segmented into sliding time windows with a fixed duration of 3 seconds, and feature calculation and index extraction are completed within each window.

[0039] S3, Emotional Load Assessment.

[0040] Specifically, in this embodiment, the following physiological characteristic indicators are calculated within each 3-second sliding time window: ΔRMSSD: The magnitude of change in heart rate variability (RMSSD) relative to the user's physiological baseline within the current time window.

[0041] ΔHF: The magnitude of change of the high-frequency component of heart rate variability (HF) relative to the physiological baseline within the current time window.

[0042] SCL slope: The slope of the change of the slow-varying component of the electrodermal signal within this time window.

[0043] The above ΔRMSSD, ΔHF, SCL slope and HRI are normalized and comprehensively calculated according to the preset weight relationship to construct the physiological arousal index BioArousal (value range 0-100) to characterize the arousal level of the user's autonomic nervous system.

[0044] Then, based on the "gaze interruption ratio" (the proportion of time the user's gaze deviates from the target stimulus within a preset time window), "backward movement speed" (the displacement speed of the user in the virtual reality environment along the direction away from the stimulus), and "number of rapid head shifts" (the number of times the user's head posture changes beyond a preset angular velocity threshold per unit time) extracted from VR behavioral data, these behavioral features are normalized and weighted to construct the behavioral avoidance index AvoidScore (with a value range of 0-100).

[0045] S4, Emotional Load Assessment In this embodiment, subjective self-assessment SUDs scores, objective physiological responses (BioArousal), and behavioral avoidance behaviors (AvoidScore) are fused from multiple dimensions to generate the emotional load index EmoLoad. The specific steps are as follows: The numerical characteristics of the BioArousal index, the AvoidScore index, and the subjective assessment data were normalized to map them to a uniform numerical range (0-100).

[0046] The normalized features are weighted based on preset weight parameters to obtain their respective weighted feature values. These weight parameters can be individually adjusted according to clinical treatment needs and different anxiety types.

[0047] Based on preset fusion rules (such as a weighted Bayesian scoring mechanism), the weighted features are comprehensively calculated to generate the emotional load index EmoLoad.

[0048] The user's current emotional state is determined by comparing the EmoLoad metric with a preset emotional load threshold range. Tolerable zone (green): EmoLoad is at a low to moderate level, and users can adapt to the current exposure intensity; Edge overload zone (yellow): EmoLoad is approaching the upper limit of the threshold, and users are experiencing a slight overload reaction; Overload zone (red): EmoLoad exceeds the threshold, and users experience significant anxiety overload; Low load zone (light green): EmoLoad remains at a low level, and the current exposure intensity is insufficient.

[0049] S5, Decision Output and Dynamic Difficulty Adjustment.

[0050] Specifically, in this embodiment, based on the determination result of the EmoLoad index, the system automatically adjusts the difficulty level of virtual reality exposure therapy to form an individualized and dynamic closed-loop exposure control. When EmoLoad enters the edge overload zone or the overload zone, the system automatically performs a real-time difficulty reduction operation, including but not limited to: Reduce the number of stimuli (e.g., reduce the number of fear stimuli, reduce highly stimuli, slow down the approach speed of stimuli); Blurring visual stimuli (reducing the clarity of scene textures and weakening visual impact); Add safety cues (increase ambient brightness, add distance markers, display safe zone prompts); Briefly pause to enhance the context (play breathing guidance sounds, slow down scene action speed); Delay the triggering of the next situational event (extend the current adaptation time, and wait for the emotional load to subside).

[0051] When EmoLoad remains in the low load zone, the system automatically performs real-time difficulty adjustment operations, including but not limited to: Increase the number of stimuli (increase the number of fear stimuli, expand the scope of stimuli coverage); Shorten the stimulus distance (bring the virtual distance between the user and the stimulus); Enhance scene immersion (increase the intensity of multi-sensory stimulation such as vision and hearing); Increase the complexity of the interaction (set up more task interaction steps, increase the response requirements).

[0052] Through the above dynamic regulation, it is ensured that the user's real-time emotional load is always guided within the treatment target range (tolerable range), which not only guarantees the treatment effect but also avoids treatment interruption due to excessive stimulation, thus achieving individualized and precise exposure therapy.

[0053] In one embodiment, the operation flow of the virtual reality exposure therapy system based on multimodal data feedback according to the present invention is as follows: Start the system, complete the connection and calibration of VR devices and physiological data acquisition devices, and establish a unified time reference; The device is worn by the user and the user enters a preset initial exposure scenario to collect the user's physiological baseline data (RMSSD, HF, SCL, etc.). Initiate exposure therapy, simultaneously collect multimodal data (physiological signals, behavioral interaction data), and receive user subjective assessment data; Perform time alignment, sliding window processing, and feature extraction on multimodal data, and calculate BioArousal, AvoidScore, and EmoLoad; Based on EmoLoad, the emotional state is determined and the difficulty is automatically adjusted. The system continuously updates the user's individual emotional profile in real time, and cycles through data collection, analysis, and regulation until the treatment ends. After treatment, a quantitative treatment report is generated, including emotional load change curves, key indicator trends, and treatment effect assessments, providing clinical reference for medical staff.

[0054] Figure 2 This is a system framework diagram of a virtual reality exposure therapy system based on multimodal data feedback according to the present invention. Figure 2 As shown, the present invention discloses a virtual reality exposure therapy system based on multimodal data feedback, comprising: Virtual reality interaction module 1 is used to present the exposed scene and collect user behavior and interaction data; Multimodal data acquisition module 2 is used to acquire the user's physiological signal data; Subjective assessment input module 3 is used to receive subjective assessment data from users during the exposure process; Data synchronization and processing module 4 is used to perform time alignment and sliding window processing on the physiological signal data, behavioral interaction data and subjective evaluation data; Emotion modeling module 5 is used to construct individualized emotion profiles and generate the emotion load index EmoLoad; The decision-making and control module 6 is used to automatically adjust the difficulty level of the virtual reality exposure scene based on the emotional load index EmoLoad.

[0055] Although embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A virtual reality exposure therapy method based on multimodal data feedback, characterized in that, The method of presenting exposure scenarios and collecting user data through virtual reality devices includes the following closed-loop steps: S1, Synchronous Acquisition and Processing of Multimodal Data: During virtual reality exposure, the system simultaneously collects users' physiological signal data and behavioral interaction data, and receives subjective evaluation data input by users. S2, Real-time Construction and Update of Personalized Emotional Profiles: Based on the physiological signal data, behavioral interaction data, and subjective evaluation data, an emotional feature model is constructed to characterize the user's emotional state. S3, Emotional Load Assessment: Based on the emotional feature model, an emotional load index, EmoLoad, is generated to characterize the user's emotional load state during virtual reality exposure. S4, Decision Output and Dynamic Difficulty Adjustment: Based on the determination result of the EmoLoad index relative to the preset emotional load range, the difficulty level of virtual reality exposure therapy is automatically adjusted so that the user's real-time emotional load is guided to the treatment target range, thereby forming an individualized and dynamic closed-loop exposure control.

2. The virtual reality exposure therapy method based on multimodal data feedback as described in claim 1, characterized in that, The physiological signal data, behavioral interaction data, and subjective evaluation data are timestamped based on a unified system time base, and after time alignment, they are segmented into segments with a fixed-length sliding time window, wherein the duration of the sliding time window is 3 seconds.

3. The virtual reality exposure therapy method based on multimodal data feedback as described in claim 2, characterized in that, Within each of the sliding time windows, the physiological signal data are processed to calculate the following physiological characteristic indicators: ΔRMSSD is used to represent the magnitude of change in the heart rate variability index relative to the user's physiological baseline within the current time window; ΔHF is used to represent the magnitude of change of the high-frequency component of heart rate variability relative to the physiological baseline within the current time window; SCL slope is used to represent the slope of the change of the slow-varying component of the skin conductance signal within the time window. The Heart Rate Response Index (HRI) is used to indicate the degree of deviation of the current heart rate level from the physiological baseline.

4. The virtual reality exposure therapy method based on multimodal data feedback as described in claim 3, characterized in that, The ΔRMSSD, ΔHF, SCL slope, and HRI are normalized and comprehensively calculated according to a preset weight relationship to construct the BioArousal index, which is used to characterize the arousal level of the user's autonomic nervous system.

5. The virtual reality exposure therapy method based on multimodal data feedback as described in claim 1, characterized in that, Based on behavioral interaction data in a virtual reality environment, the following behavioral features are extracted: The gaze interruption ratio represents the percentage of time a user's gaze deviates from the target stimulus within a preset time window. Backward motion speed, which represents the user's displacement speed in the virtual reality environment along the direction away from the stimulus; The number of rapid head shifts indicates the number of times the user's head posture changes beyond a preset angular velocity threshold per unit time. The behavioral characteristics are then normalized and weighted to construct the AvoidScore, a behavioral avoidance index.

6. The virtual reality exposure therapy method based on multimodal data feedback as described in claim 1, characterized in that, The subjective assessment data consists of the subjective pain or emotional intensity ratings input by users during virtual reality exposure. These ratings are divided into multiple level ranges according to a preset scale and converted into corresponding numerical features for the calculation of emotional load indicators.

7. The virtual reality exposure therapy method based on multimodal data feedback as described in any one of claims 1 to 7, characterized in that, The generation of the emotional load index EmoLoad includes the following steps: The numerical characteristics of the BioArousal index, AvoidScore index, and subjective evaluation data are normalized to map them to a uniform numerical range. Based on preset weight parameters, the normalized features are weighted and calculated to obtain their respective weighted feature values. Based on the weighted feature values, the weighted features are comprehensively calculated according to the preset fusion rules to generate the emotional load index EmoLoad; The emotional load index EmoLoad is compared with a preset emotional load threshold range to determine whether the user's current emotional state is in the tolerable zone, the edge overload zone, the overload zone, or the low load zone.

8. The virtual reality exposure therapy method based on multimodal data feedback as described in claim 7, characterized in that, When the EmoLoad indicator is detected to enter the preset edge zone or overload zone, the system automatically performs real-time difficulty adjustment operations, including reducing the number of stimuli, blurring visual stimuli, increasing the distance between stimuli, increasing safety cues, briefly pausing the enhanced context, and delaying the division of the next contextual event, thereby forming an individualized and dynamic closed-loop exposure control process.

9. A virtual reality exposure therapy method based on multimodal data feedback as described in claim 7, characterized in that, When the EmoLoad indicator is detected to be continuously in the low load zone, the system automatically performs real-time difficulty adjustment operations, including increasing the number of stimuli, shortening the distance between stimuli, and increasing the immersion of the scene or the complexity of the interaction, thereby forming an individualized and dynamic closed-loop exposure control process.

10. A virtual reality exposure therapy system based on multimodal data feedback, characterized in that, include: The virtual reality interaction module is used to present exposed scenes and collect user behavior and interaction data; A multimodal data acquisition module is used to collect users' physiological signal data; The subjective evaluation input module is used to receive subjective evaluation data from users during the exposure process; The data synchronization and processing module is used to perform time alignment and sliding window processing on the physiological signal data, behavioral interaction data and subjective evaluation data; The emotion modeling module is used to build individualized emotion profiles and generate the emotion load index EmoLoad; The decision-making and control module is used to automatically adjust the difficulty level of virtual reality exposure scenarios based on the EmoLoad index.

Images