Wound memory integration evaluation and treatment system

Abstract

The invention provides a trauma memory integration evaluation and treatment system. Comprising the following steps: acquiring patient description information, generating a trauma event scene based on the description information, collecting multi-modal data of a patient in the scene, identifying a trauma recall activation state by utilizing the multi-modal data, combining the activation state with individual psychological characteristics and database contents to generate an intervention scheme, and circularly updating the scene and the scheme until a termination condition is met. The importance of the multi-modal features is determined through a machine learning model, a trauma recall activation score is generated through weighted fusion, and quantitative and objective evaluation of the activation state is achieved; and a personalized intervention scheme is dynamically retrieved or generated in combination with psychological characteristics of the patient, so that the pertinence and effectiveness of treatment are improved. The technical problems that existing trauma recall activation evaluation depends on subjective judgment, and an intervention strategy lacks individuation and dynamic optimization can be solved.

Description

Technical Field

[0001] This invention relates to the field of psychotherapy technology, specifically to a trauma memory integration assessment and treatment system. Background Technology

[0002] Post-traumatic stress disorder (PTSD) is a common mental health problem that arises after experiencing or witnessing a severe traumatic event. Its core symptoms are triggered by recurring memories of the traumatic event, which significantly impact the individual's daily life, mental health, and social functioning. Current treatments for PTSD include psychotherapy such as cognitive behavioral therapy, exposure therapy, mindfulness therapy, and eye movement desensitization reprocessing, as well as medication. However, psychotherapy is highly dependent on the experience and skills of therapists, treatment resources are limited, and consistent treatment outcomes are difficult to guarantee. While medication can alleviate some symptoms, it has side effects and cannot fundamentally repair the traumatic memories.

[0003] Current trauma psychotherapy is largely unable to reconstruct the traumatic event scenario. Relying solely on patients' oral accounts or experiencing / reliving the traumatic event inevitably limits the therapist's empathy and can lead to secondary trauma and unnecessary costs for patients. Therefore, how to utilize virtual scenario technology to realistically reconstruct the traumatic event scenario in a controlled and safe environment, and combine this with objective indicators to conduct real-time assessment and intervention of the patient's trauma activation state, thereby improving treatment effectiveness, reducing the risk of secondary trauma, and lessening reliance on highly skilled therapists, is a pressing technical problem that needs to be solved in this field. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention proposes a trauma memory integration assessment and treatment system to solve the problems of existing technologies, such as trauma memory activation assessment relying on subjective judgment and intervention strategies lacking personalization and dynamic optimization techniques.

[0005] The technical solution adopted in this invention is a trauma memory integration assessment and treatment system, comprising: a multimodal data acquisition module for acquiring the patient's descriptive information; A trauma scene reconstruction module is used to generate a trauma event scene based on the description information; A trauma memory recognition module is used to collect multimodal data of the patient in the trauma event scenario; The trauma memory analysis module is used to identify the activation state of trauma memories by using the multimodal data. The personalized intervention strategy generation module is used to combine the activation state of the traumatic memory with individual psychological characteristics and database content to generate an intervention plan.

[0006] Furthermore, the personalized intervention strategy generation module also includes: inputting the intervention plan as descriptive information into the trauma scene reconstruction module to update the trauma event scene until the clinical assessment termination conditions are met, and then terminating the treatment.

[0007] Furthermore, the multimodal data acquisition module includes: acquiring the patient's basic personal information, personal medical history, and an overview of the traumatic events through multimodal sensors; The aforementioned basic personal information, medical history, and traumatic events were compiled into a text message. The text information is semantically understood and analyzed to extract descriptive information related to the traumatic event; The descriptive information includes keywords, event context, event logic, and sentiment.

[0008] Furthermore, the trauma scene reconstruction module includes: generating at least one trauma event scene including images, videos and text content in the AI ​​large model based on the description information; The traumatic event scene is a panoramic video generated from a first-person perspective.

[0009] Furthermore, in the trauma memory recognition module, the multimodal data includes physiological indicators, emotional response levels, and standard psychological scales; The physiological indicators include heart rate, skin conductance, respiratory rate, body temperature, blood oxygen saturation, and pulse. The emotion data includes facial expression images, body postures, and voice features; The standard psychological scale is completed by patients through a visual interface and voice guidance.

[0010] Furthermore, the trauma memory analysis module includes: preprocessing, feature extraction, and standardization of the physiological indicators, emotional response levels, and standard psychological scales to obtain physiological feature vectors, emotional feature vectors, and psychological score vectors, respectively. The importance of each feature is determined based on training data with real activation annotations; The scores of each feature within the same modality are weighted according to the importance of the features to obtain the weighted comprehensive score of the modality; The importance of each modality is normalized into modality weights, and the weighted composite scores of each modality are weighted and fused to obtain the patient's trauma memory activation score.

[0011] Furthermore, the importance of each feature is determined based on training data with real activation annotations, including: constructing a training set based on the physiological feature vector, emotional feature vector, and psychological rating vector; The training set is trained using an interpretable machine learning model to obtain an importance score for each input feature; The interpretable machine learning model is one of the following: random forest, gradient boosting tree, or attention-based neural network.

[0012] Furthermore, the scores of each feature within the same modality are weighted according to the importance of the feature to obtain the weighted comprehensive score of the modality, including: normalizing the importance scores of each feature within the same modality into in-modality weights according to the proportion; For the feature set within the same modality The intramodal weights can be expressed as: in, Indicates the intramodal weights. This indicates the importance score of the feature currently being processed; The importance scores of all features within the same modality are weighted according to the weights within the modality to synthesize a weighted composite score for that modality.

[0013] Furthermore, the importance of each modality is normalized into modal weights, and the weighted composite scores of each modality are weighted and fused to obtain the patient's trauma memory activation score, including: calculating the modal weight of each modality according to the cumulative importance calculation formula: in, Modal weights representing psychological rating modalities. Let represent the importance score of the j-th feature in the psychological rating modality, and Q represent the feature set of the psychological rating modality. Modal weights representing emotional modalities Let Y represent the importance score of the j-th feature in the emotion modality, and let Y represent the feature set of the emotion modality. Modal weights representing physiological modalities Let X represent the importance score of the j-th feature in the physiological modality, and let X represent the feature set of the physiological modality. The modal weights are obtained by normalizing the modal weights: in, Indicates the modal weights of psychological ratings. Represents the modality weights of sentiment. Represents physiological modality weights; Based on the modality weights, a weighted composite score of all modalities is fused to generate a traumatic memory activation score: Where A represents the traumatic memory activation score, , , These are the weighted composite scores for the psychological rating modality, the emotional modality, and the physiological modality, respectively.

[0014] Preferably, the personalized intervention strategy generation module includes: determining the patient's activation level based on the trauma memory activation score; Extract at least one of the following from the individual psychological characteristics: personality traits, coping styles, attachment styles, previous psychological assessment results, and psychological scale scores; The intervention strategy template that matches the activation level and the individual psychological characteristics is retrieved from the intervention strategy database. The trained intervention generation model is then invoked to generate an intervention plan based on the intervention strategy template, the activation level, and the individual psychological characteristics.

[0015] As can be seen from the above technical solution, the beneficial technical effects of the present invention are as follows: 1. By collecting multimodal data and using interpretable machine learning models to determine feature importance, and by using intramodal weighting and intermodal weight fusion methods to generate trauma memory activation scores, a quantitative and objective assessment of the activation state of trauma memory is achieved, reducing the uncertainty of subjective human judgment.

[0016] 2. By combining the trauma memory activation score with the patient's individual psychological characteristics, intervention plans are dynamically generated through intervention plan database retrieval or training models, and continuously updated and optimized in the treatment cycle to ensure that the intervention plan can adapt to the patient's psychological state changes in real time, thereby improving the pertinence and effectiveness of treatment.

[0017] 3. Immersive first-person trauma event scenarios generated by AI large models can replace traditional on-site or imagined exposure, reducing secondary psychological impact on patients during treatment, while also reducing the high dependence on therapists' empathy and experience, and improving the accessibility and consistency of treatment. Attached Figure Description

[0018] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0019] Figure 1 This is a system structure diagram of an embodiment of the present invention; Figure 2 This is a trauma assessment mapping table according to an embodiment of the present invention; Figure 3 This is a system flowchart of an embodiment of the present invention. Detailed Implementation

[0020] The embodiments of the technical solution of the present invention will now be described in detail with reference to the accompanying drawings. These embodiments are merely illustrative of the technical solution of the present invention and are therefore intended to limit the scope of protection of the present invention.

[0021] It should be noted that, unless otherwise stated, the technical or scientific terms used in this application should have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains.

[0022] This embodiment provides a trauma memory integration assessment and treatment system. The working principle of Embodiment 1 is explained in detail below: The system structure diagram of this embodiment is as follows: Figure 1 As shown, it includes: The multimodal data acquisition module is used to obtain descriptive information about the patient.

[0023] The trauma scene reconstruction module is used to generate a trauma event scene based on the described information.

[0024] The trauma memory recognition module is used to collect multimodal data of the patient in the trauma event scenario.

[0025] The trauma memory analysis module is used to identify the activation state of trauma memories by using the multimodal data. A personalized intervention strategy generation module is used to combine the activation state of the traumatic memory with individual psychological characteristics and database content to generate an intervention plan. The personalized intervention strategy generation module further includes: inputting the intervention plan as descriptive information into the trauma scene reconstruction module to update the traumatic event scene until the clinical assessment termination criteria are met, at which point treatment is terminated.

[0026] Furthermore, by setting physiological indicator thresholds (HR (heart rate) > 150 beats / min), the system automatically pauses scenario generation and triggers a treatment termination alarm, providing clear reminders and enabling direct intervention by a psychological counselor or referral to a hospital; the person seeking help can terminate the treatment themselves via voice commands (such as "stop") or an emergency button.

[0027] The multimodal data acquisition module is equipped with various sensors, including but not limited to cameras, microphones, and physiological sensors (including electrocardiogram, skin conductance sensors, skin temperature, respiration, and posture sensors). The camera captures visual information such as facial expressions and body movements of the person seeking psychological help; the microphone collects audio information such as the person's speech content, tone, and speech rate; and the physiological sensors monitor changes in physiological indicators such as heart rate, skin conductance, skin temperature, and pulse in real time during the process of recalling traumatic events or interacting with the multimodal model. These physiological indicators can effectively reflect the emotional responses of the person seeking psychological help. To comprehensively collect data related to traumatic memories, it is necessary to integrate multiple types of data, including physiological, behavioral, and linguistic data. This includes collecting physiological data such as heart rate and skin conductance. For example, a health bracelet can continuously monitor heart rate and skin conductance activity, accurately capturing an individual's physiological stress response when recalling a traumatic event. Furthermore, cameras and microphones can be used to record an individual's facial expressions, body movements, and verbal expressions, including speech rate, tone of voice, and gestures.

[0028] In clinical settings, high-definition cameras can be installed at multiple angles and from the front of the therapy room to ensure complete recording of the client's full-body movements during therapy. Simultaneously, professional recording equipment should be deployed to clearly capture the conversations between the client and therapist. For wearable devices, users should be given detailed instructions on how to wear them and any precautions beforehand to ensure the accuracy and stability of the data collection.

[0029] Data collected from different devices is synchronized to ensure time consistency. During the acquisition process, the data undergoes preliminary filtering to remove high-frequency noise and improve data quality.

[0030] The trauma memory recognition module combines facial expression recognition algorithms, limb movement and body posture analysis algorithms, and physiological signal analysis models such as electrocardiogram, skin impedance, skin temperature, and respiration to comprehensively determine whether the person seeking psychological help is currently in a state of trauma memory arousal, and assess the severity of the trauma memory and its impact on the person's emotions. For example, by analyzing facial expressions such as fear, sadness, distress, and smiles, combined with cues in speech such as voice trembling, pauses, high-pitched tone, or high-frequency repetition, as well as changes in heart rate, skin conductance, skin temperature, and respiratory response, the module quantifies the intensity of the trauma memory's stimulation of the person seeking psychological help.

[0031] In this embodiment, the system flowchart is further shown below. Figure 3 As shown, the multimodal data acquisition module includes: the system guides the psychological helper to recall their traumatic events through voice inquiry and text dialogue, and obtains the patient's descriptive information based on prompt word templates.

[0032] Obtain the patient's descriptive information based on the prompt template, including basic personal information, personal medical history, and an overview of the traumatic event.

[0033] The system collects patients' basic personal information, medical history, and trauma event overview by combining voice recognition input with text input from patients, and then organizes this information into text.

[0034] The text information, such as the content and details describing the traumatic event, is semantically understood and analyzed to extract descriptive information related to the traumatic event, namely, the contextual description text.

[0035] The emotional description information includes keywords, event context, event logic, and emotional tendency; the emotional tendency includes emotional experiences such as pain, fear, and tension.

[0036] This embodiment provides a commonly used prompting word template for psychological help seekers: (1) What is your current psychological distress? (2) When, where, what object or person, and your status (age, gender, education, self-evaluation: excellent, average, or failure)? (3) What happened and your mood? Choose your feeling: fear, surprise, or anger? (4) What impressed you the most? Please describe the details in five words (ten characters).

[0037] In this embodiment, the trauma scene reconstruction module further includes: generating at least one trauma event scene including images, videos and text content in the AI ​​large model based on the description information; The traumatic event scene is a panoramic video generated from a first-person perspective.

[0038] The generated trauma event scenario is designed to maximize the emotional response of the patient, including enhancing images, sounds, or environmental features that are relevant to the core details of the trauma.

[0039] Multimodal AI technologies, including text-to-image, text-to-video, and image-to-video, are applied to reconstruct traumatic scenarios. Based on natural language processing, image generation, and video generation technologies using large language models, semantic understanding and analysis are performed on the text information input by psychological help seekers to identify keywords, event logic, and emotional tendencies related to the traumatic event.

[0040] Based on large-scale model image and video generation technologies, the system processes relevant images such as portrait photos of the person seeking help, detailed photos of specific environments and scenes, etc., to generate similar trauma scenes, so as to highly restore the scene where the traumatic event occurred. At the same time, it can also generate therapeutic scenes that are conducive to psychological rehabilitation according to the needs of psychotherapy.

[0041] In this embodiment, the multimodal data includes physiological indicators, emotional response levels, and standard psychological scales.

[0042] The physiological indicators include heart rate, skin conductance, respiratory rate, body temperature, blood oxygen saturation, and pulse.

[0043] The emotion data includes facial expression images, body postures, and voice features.

[0044] The standard psychological scales are completed by patients through a visual interface and voice guidance. The scales include: SAS Self-Rating Anxiety Scale, PCL-17, SDS Self-Rating Depression Scale, Anxiety Trait Scale, SCL-90 Scale, Dissociation Scale, Pittsburgh Sleep Scale, Daily Living Scale, Subjective Distress Scale, Personality Trait Scale, etc., totaling fifteen scales.

[0045] Furthermore, the trauma memory recognition module combines facial expression recognition algorithms, limb movement and body posture analysis algorithms, and physiological signal analysis models such as electrocardiogram, skin impedance, skin temperature, and respiration to comprehensively determine whether the psychological help seeker is currently in a state of trauma memory arousal, and assess the severity of the trauma memory and its impact on the psychological helper's emotions. This includes analyzing features such as fear, sadness, distress, and smiles in facial expressions, combining cues such as voice trembling, pauses, high-pitched tone, or high-frequency repetition in speech, and the magnitude of changes in heart rate, skin conductance, skin temperature, and respiratory response to quantify the intensity of the psychological stimulation of the psychological helper by analyzing features such as fear, sadness, distress, and smiles in facial expressions, combining cues such as voice trembling, pauses, high-pitched tone, or high-frequency repetition in speech, and the magnitude of changes in heart rate, skin conductance, skin temperature, and respiratory response.

[0046] Furthermore, in this embodiment, the trauma memory analysis module includes: preprocessing, feature extraction, and standardization of the physiological indicators, emotional response levels, and standard psychological scales to obtain physiological feature vectors, emotional feature vectors, and psychological score vectors, respectively.

[0047] The importance of each feature is determined based on training data with real activation annotations.

[0048] The scores of each feature within the same modality are weighted according to the importance of the feature to obtain the weighted comprehensive score of that modality.

[0049] The importance of each modality is normalized into modality weights, and the weighted composite scores of each modality are weighted and fused to obtain the patient's trauma memory activation score.

[0050] The physiological indicators, emotional response levels, and standard psychological scales are preprocessed, feature extracted, and standardized to obtain physiological feature vectors, emotional feature vectors, and psychological score vectors, including: For the aforementioned standard psychological scale, each item of the standard psychological scale is scored and normalized according to the official scoring criteria to generate a psychological score vector representing the feature set of that feature: in, Represents the psychological rating vector. This represents the score of the nth scale.

[0051] For the physiological indicators and emotional response levels, each feature is mapped to a score of 0–10 according to a preset trauma assessment mapping table, thus mapping the physiological indicators and emotional response levels into a score. The trauma assessment mapping table is as follows: Figure 2 As shown.

[0052] The physiological indicators and emotional response levels are mapped to physiological feature vectors. and emotion feature vector .

[0053] In this embodiment, further, the importance of each feature is determined based on training data with real activation annotations, including: constructing a training set based on the physiological feature vector, emotional feature vector, and psychological rating vector.

[0054] The training set is trained using an interpretable machine learning model to obtain an importance score for each input feature.

[0055] The interpretable machine learning model is one of the following: random forest, gradient boosting tree, or attention-based neural network.

[0056] In this embodiment, further, the scores of each feature within the same modality are weighted according to the importance of the feature to obtain the weighted comprehensive score of the modality, including: normalizing the importance scores of each feature within the same modality into in-modality weights according to a proportion.

[0057] For physiological modalities: the set of features within the same modality. The intramodal weights can be expressed as: in, This represents the in-mode weights of the current mode, i.e., the physiological mode. This indicates the importance score of the feature currently being processed.

[0058] The importance scores of all features within the same modality are weighted according to the weights within that modality to synthesize a weighted comprehensive score for that modality: in, This represents the weighted composite score of physiological modalities. This represents the in-modal weights of the premodal, i.e., physiological modality. This represents the standardized score of the i-th feature in the physiological modality.

[0059] Similarly, the intramodal importance score is applied to the emotional feature vector Y and the psychological rating vector Q. , Weighted synthetic sentiment modality weighted importance score. Weighted importance scores of psychological scales : Where Y represents the emotion feature vector, This represents the weighted composite score of the modalities. The in-modal weights represent the current modality, and Q represents the psychological rating vector. This represents the weighted composite score of psychological assessments.

[0060] In this embodiment, the importance of each modality is further normalized into modality weights, and the weighted composite scores of each modality are weighted and fused to obtain the patient's trauma memory activation score, including: The importance of each mode is normalized into its modal weight, including: The modal weight, or cumulative importance, for each mode is calculated using the cumulative importance calculation formula: in, The modal weights, or total importance, represent the modalities of psychological rating modalities. Let represent the importance score of the j-th feature in the psychological rating modality, and Q represent the feature set of the psychological rating modality. Modal weights representing emotional modalities Let Y represent the importance score of the j-th feature in the emotion modality, and let Y represent the feature set of the emotion modality. Modal weights representing physiological modalities Let X represent the importance score of the j-th feature in the physiological modality, and let X represent the feature set of the physiological modality.

[0061] The modal weights are obtained by normalizing the modal weights: in, Indicates the modal weights of psychological ratings. Represents the modality weights of sentiment. This represents the physiological modality weights.

[0062] Based on the modality weights, a weighted composite score of all modalities is fused to generate a traumatic memory activation score: Where A represents the traumatic memory activation score, , , The scores are weighted composite scores for physiological, emotional, and psychological modalities, respectively.

[0063] This method enables precise multimodal identification and analysis of traumatic memories. Compared with traditional single assessment methods, it provides a more comprehensive and accurate understanding of the trauma state of those seeking psychological help, thus providing a solid foundation for personalized intervention.

[0064] In this embodiment, the personalized intervention strategy generation module further includes: determining the patient's activation level based on the trauma memory activation score.

[0065] Personality traits, coping styles, attachment styles, previous psychological assessment results, and psychological scale scores are extracted from the individual's psychological characteristics.

[0066] The intervention strategy template that matches the activation level and the individual psychological characteristics is retrieved from the intervention strategy database. The trained intervention generation model is then invoked to generate an intervention plan based on the intervention strategy template, the activation level, and the individual psychological characteristics.

[0067] The intervention plan includes at least one of the following: intervention type, intervention sequence, intervention intensity, intervention duration and frequency, scenario exposure settings and coping strategies. The trained intervention generation model is a decision tree model, a random forest model, a neural network model, a reinforcement learning-based policy model, or a combination thereof.

[0068] Based on the results of trauma memory identification and analysis, personalized psychotherapy is implemented, utilizing a pre-trained personalized intervention model to generate unique intervention strategies for each client. This model integrates psychotherapeutic theories and methods such as cognitive therapy, eye movement desensitization reprocessing, exposure desensitization cognition, and mindfulness therapy. In this embodiment, analysis results show that the client's traumatic memories mainly lead to cognitive biases (such as excessive self-blame). The multimodal model will generate dialogue content and practice tasks to guide the client in cognitive reconstruction. If the client exhibits strong emotional reactions, the multimodal model will provide emotion regulation strategies such as relaxation training and breathing regulation, presented to the client in various forms such as voice, images, or videos. The personalized intervention strategy generation module can tailor treatment plans for each psychological helper to match their unique traumatic experiences and psychological characteristics, thereby improving the pertinence and effectiveness of treatment and making up for the shortcomings of the one-size-fits-all approach in traditional psychotherapy.

[0069] Furthermore, some PTSD sufferers have reservations about communicating with real people. Using a patient-oriented multimodal model can alleviate their psychological stress and achieve better results. Traditional psychotherapy, limited by therapists' time and energy, has a relatively fixed treatment cycle and frequency, making it difficult to fully meet the needs of sufferers seeking treatment at any time. While emphasizing personalization, it is limited by therapists' personal experience and methodologies, making it difficult to perfectly adapt to each sufferer. In this embodiment, the trauma memory integration learning platform, reconstructed through multimodal AI scenarios, can respond to sufferers' needs at any time, providing 24 / 7 uninterrupted service. In particular, the personalized reconstruction of trauma scenarios allows for rapid analysis of individual sufferer characteristics, the development of highly personalized treatment plans, and dynamic adjustments based on the sufferer's responses during treatment, achieving more efficient and precise personalized therapy.

[0070] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A trauma memory integration assessment and treatment system, characterized in that, include: A multimodal data acquisition module is used to obtain descriptive information about the patient; A trauma scene reconstruction module is used to generate a trauma event scene based on the description information; A trauma memory recognition module is used to collect multimodal data of the patient in the trauma event scenario; The trauma memory analysis module is used to identify the activation state of trauma memories by using the multimodal data. The personalized intervention strategy generation module is used to combine the activation state of the traumatic memory with individual psychological characteristics and database content to generate an intervention plan.

2. The trauma memory integration assessment and treatment system according to claim 1, characterized in that, The personalized intervention strategy generation module further includes: inputting the intervention plan as descriptive information into the trauma scene reconstruction module to update the trauma event scene until the clinical assessment termination conditions are met, and then terminating the treatment.

3. The trauma memory integration assessment and treatment system according to claim 1, characterized in that, The multimodal data acquisition module includes: The patient's basic personal information, medical history, and traumatic event overview are collected using multimodal sensors. The aforementioned basic personal information, medical history, and traumatic events were compiled into a text message. The text information is semantically understood and analyzed to extract descriptive information related to the traumatic event; The descriptive information includes keywords, event context, event logic, and sentiment.

4. The trauma memory integration assessment and treatment system according to claim 3, characterized in that, The trauma scene reconstruction module includes: Based on the described information, at least one trauma event scenario including images, videos, and text content is generated in the AI ​​large model; The traumatic event scene is a panoramic video generated from a first-person perspective.

5. The trauma memory integration assessment and treatment system according to claim 1, characterized in that, In the trauma memory recognition module, the multimodal data includes physiological indicators, emotional response levels, and standard psychological scales; The physiological indicators include heart rate, skin conductance, respiratory rate, body temperature, blood oxygen saturation, and pulse. The emotion data includes facial expression images, body postures, and voice features; The standard psychological scale is completed by patients through a visual interface and voice guidance.

6. The trauma memory integration assessment and treatment system according to claim 5, characterized in that, The trauma memory analysis module includes: The physiological indicators, emotional response levels, and standard psychological scales are preprocessed, feature extracted, and standardized to obtain physiological feature vectors, emotional feature vectors, and psychological score vectors, respectively. The importance of each feature is determined based on training data with real activation annotations; The scores of each feature within the same modality are weighted according to the importance of the features to obtain the weighted comprehensive score of the modality; The importance of each modality is normalized into modality weights, and the weighted composite scores of each modality are weighted and fused to obtain the patient's trauma memory activation score.

7. The trauma memory integration assessment and treatment system according to claim 6, characterized in that, The importance of each feature is determined based on training data with real activation annotations, including: A training set is constructed based on the aforementioned physiological feature vector, emotional feature vector, and psychological rating vector; The training set is trained using an interpretable machine learning model to obtain an importance score for each input feature; The interpretable machine learning model is one of the following: random forest, gradient boosting tree, or attention-based neural network.

8. The trauma memory integration assessment and treatment system according to claim 7, characterized in that, The scores of each feature within the same modality are weighted according to the importance of the features to obtain a weighted comprehensive score for that modality, including: The importance scores of each feature within the same modality are proportionally normalized into in-modality weights; For the feature set within the same modality The intramodal weights can be expressed as: in, Indicates the intramodal weights. This indicates the importance score of the feature currently being processed; The importance scores of all features within the same modality are weighted according to the weights within the modality to synthesize a weighted composite score for that modality.

9. A trauma memory integration assessment and treatment system according to claim 8, characterized in that, The importance of each modality is normalized into modality weights. The weighted composite scores of each modality are then weighted and fused to obtain the patient's trauma memory activation score, including: The modal weight for each mode is calculated using the cumulative importance formula: in, Modal weights representing psychological rating modalities. Let represent the importance score of the j-th feature in the psychological rating modality, and Q represent the feature set of the psychological rating modality. Modal weights representing emotional modalities Let Y represent the importance score of the j-th feature in the emotion modality, and let Y represent the feature set of the emotion modality. Modal weights representing physiological modalities Let X represent the importance score of the j-th feature in the physiological modality, and let X represent the feature set of the physiological modality. The modal weights are obtained by normalizing the modal weights: in, Indicates the modal weights of psychological ratings. Represents the modality weights of sentiment. Represents physiological modality weights; Based on the modality weights, a weighted composite score of all modalities is fused to generate a traumatic memory activation score: Where A represents the traumatic memory activation score, , , These are the weighted composite scores for the psychological rating modality, the emotional modality, and the physiological modality, respectively.

10. A trauma memory integration assessment and treatment system according to claim 1, characterized in that, The personalized intervention strategy generation module includes: The patient's activation level is determined based on the trauma memory activation score; Extract at least one of the following from the individual psychological characteristics: personality traits, coping styles, attachment styles, previous psychological assessment results, and psychological scale scores; The intervention strategy template that matches the activation level and the individual psychological characteristics is retrieved from the intervention strategy database. The trained intervention generation model is then invoked to generate an intervention plan based on the intervention strategy template, the activation level, and the individual psychological characteristics.

Images