Method and device for recommending individualized intervention strategy for subthreshold depression based on multi-modal data
By acquiring multimodal data and training machine learning models, a subthreshold depression screening model was constructed, which solved the standardization problem of subthreshold depression identification and intervention in existing technologies and achieved efficient and accurate recommendations for individualized intervention strategies.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- PEKING UNIVERSITY SIXTH HOSPITAL
- Filing Date
- 2024-12-12
- Publication Date
- 2026-06-12
Smart Images
Figure CN122201641A_ABST
Abstract
Description
Technical Field
[0001] This disclosure generally relates to the field of artificial intelligence, and specifically to a method and apparatus for recommending individualized intervention strategies for subthreshold depression based on multimodal data. Background Technology
[0002] Physical examinations play an irreplaceable role in the early detection and prevention of diseases, but the development of mental health checkups has lagged behind and has not received sufficient attention. Studies show that approximately 15% of individuals undergoing routine physical examinations exhibit serious psychological problems, most of whom experience symptoms of depression or anxiety. Subthreshold depression, a sub-healthy mental state between health and depression, has an incidence rate of approximately 7.3% to 23.1% in the general population. This group is considered high-risk for depression; about 10% of those with subthreshold depression may progress to depression within 1 to 2 years, while 30% may develop depression within 6 years. Failure to identify and intervene in individuals with subthreshold depression in a timely manner will not only increase potential risks to individual health but also significantly burden the social healthcare system.
[0003] Currently, the identification and intervention of individuals with subthreshold depression can only rely on on-site manual diagnosis by doctors. On the one hand, the identification process is highly dependent on the professional judgment of doctors and lacks standardized assessment tools, which limits the consistency and accuracy of diagnosis. Moreover, manual diagnosis is time-consuming, especially when there are a large number of patients, making it difficult to achieve large-scale screening and affecting early detection and timely intervention. On the other hand, patients' behavior of seeking medical treatment is usually significantly delayed. In most cases, patients only actively seek medical help when their subthreshold depression worsens to mild or severe depression. This not only misses the best intervention opportunity but also increases the difficulty and complexity of treatment. Summary of the Invention
[0004] In view of the above-mentioned defects or deficiencies in the prior art, it is desirable to provide a method and device for recommending individualized intervention strategies for subthreshold depression based on multimodal data, which can realize automated screening and intervention strategy recommendation for subthreshold depression.
[0005] In a first aspect, embodiments of this application provide a method for recommending individualized intervention strategies for subthreshold depression based on multimodal data, including:
[0006] To acquire multimodal psychological data from normal individuals, patients with subthreshold depression, and patients with depression;
[0007] The multimodal psychological data is preprocessed and added to the database;
[0008] The pre-built machine learning model is trained for subthreshold depression screening based on the database, and the trained model is used as the screening model.
[0009] Receive user data to be identified, input it into the screening model for identification, and generate screening results for subthreshold depression;
[0010] Based on the screening results, determine whether the user is in a state requiring intervention; if so, extract intervention suggestions from the pre-built subthreshold depression intervention database, and generate an intervention strategy based on the intervention suggestions.
[0011] In some embodiments, the multimodal psychological data includes: scale assessment data, cognitive test data, heart rate variability data, and micro-expression videos.
[0012] In some embodiments, the preprocessing of the multimodal psychological data and its addition to the database includes:
[0013] The scale assessment data, the cognitive test data, and the heart rate variability data are extracted as numerical data; the numerical data are screened for missing, abnormal, and logically incorrect data; the screened data is then added to the database.
[0014] The micro-expression video is extracted frame by frame into numerical data, which is used as micro-expression data; the micro-expression data is input into a classification model for action classification, the output action classification results are obtained, and added to the database.
[0015] In some embodiments, the specific components include:
[0016] The micro-expression video is extracted frame by frame into numerical data, which is used as micro-expression data;
[0017] The micro-expression data is input into a classification model to classify actions, and the output action classification results are obtained.
[0018] Calculate the average Gini index for each of the action classification results;
[0019] The top N values of the average reduction in the Gini index from largest to smallest disadvantage are used to extract the action classification results corresponding to the top N values and add them to the database.
[0020] In some embodiments, the step of extracting intervention suggestions from a pre-built subthreshold depression intervention library and generating intervention strategies based on the intervention suggestions includes:
[0021] The type of subthreshold depression was determined based on the screening results;
[0022] Intervention suggestions corresponding to the aforementioned types are extracted from a pre-built subthreshold depression intervention database, and intervention strategies are generated based on these suggestions.
[0023] Secondly, embodiments of this application provide a device for recommending individualized intervention strategies for subthreshold depression based on multimodal data, including:
[0024] The modal data acquisition unit is used to acquire multimodal psychological data from normal individuals, subthreshold depression patients, and depression patients.
[0025] The data preprocessing unit is used to preprocess the multimodal psychological data and add it to the database;
[0026] The screening training unit is used to train a pre-built machine learning model for subthreshold depression screening based on the database, and obtain the trained model as the screening model.
[0027] The data recognition unit receives user data to be identified, inputs it into the screening model for recognition, and generates screening results for subthreshold depression.
[0028] An intervention recommendation unit is used to determine whether a user is in a state requiring intervention based on the screening results; if so, it extracts intervention suggestions from a pre-built subthreshold depression intervention database and generates an intervention plan based on the intervention suggestions.
[0029] In some embodiments, the intervention unit further includes: an intervention library optimization subunit, used to acquire user follow-up data after intervention; and to optimize the intervention recommendations in the subthreshold depression intervention library based on the user follow-up data.
[0030] Thirdly, embodiments of this application provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method described in embodiments of this application.
[0031] Fourthly, embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in embodiments of this application.
[0032] Fifthly, embodiments of this application provide a computer program product, including a computer program, characterized in that, when the computer program is executed by a processor, it implements the method described in embodiments of this application.
[0033] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description
[0034] Other features, objects, and advantages of this application will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:
[0035] Figure 1 The diagram illustrates a flowchart of a method for recommending individualized intervention strategies for subthreshold depression based on multimodal data, according to an embodiment of this application.
[0036] Figure 2 An exemplary structural block diagram of a screening and intervention device for subthreshold depression provided in an embodiment of this application is shown;
[0037] Figure 3 An exemplary structural block diagram of a screening and intervention system for subthreshold depression provided in an embodiment of this application is shown. Detailed Implementation
[0038] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.
[0039] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.
[0040] Subthreshold depression, also known as mild depression or subclinical depression, is a mental state that falls between normal mood swings and clinical depression. Although its symptoms do not meet the diagnostic criteria for clinical depression, subthreshold depression still has a significant negative impact on an individual's daily life and functioning. This state is characterized by the presence of depressive symptoms, but the severity and duration of these symptoms are insufficient to meet the formal diagnostic criteria for depression in the International Classification of Diseases (ICD-10) or the Diagnostic and Statistical Manual of Mental Disorders (DSM-5). Individuals with subthreshold depression often experience symptoms such as depressed mood, loss of interest, fatigue and lack of energy, sleep disturbances, and impaired cognitive function. These symptoms typically last for weeks to months, but do not meet the minimum duration requirement of at least two weeks for clinical depression. These symptoms may occur periodically, causing recurring distress and significantly impacting an individual's daily life and functioning.
[0041] To further illustrate the technical solutions provided in the embodiments of this application, a detailed description is provided below in conjunction with the accompanying drawings and specific implementation methods. Although the embodiments of this application provide method operation instruction steps as shown in the following embodiments or drawings, the method may include more or fewer operation instruction steps based on conventional or non-creative effort. In steps where there is no logically necessary causal relationship, the execution order of these steps is not limited to the execution order provided in the embodiments of this application. In actual processing or when the device executes the method, it may be executed sequentially or in parallel according to the method shown in the embodiments or drawings.
[0042] It should be noted that the acquisition or use of data in the embodiments of this application requires the user's consent. The relevant data can only be obtained after the user's authorization, and the acquisition or use of the data complies with relevant laws and regulations.
[0043] Please refer to Figure 1 , Figure 1 This illustration shows a flowchart of a method for recommending individualized intervention strategies for subthreshold depression based on multimodal data, according to an embodiment of this application. Figure 1 As shown, the method includes:
[0044] Step 101: Obtain multimodal psychological data from normal individuals, patients with subthreshold depression, and patients with depression.
[0045] Step 102: Preprocess the multimodal psychological data and add it to the database.
[0046] Step 103: Train the pre-built machine learning model for subthreshold depression screening based on the database, and obtain the trained model as the screening model.
[0047] Step 104: Receive the user data to be identified, input it into the screening model for identification, and generate screening results for subthreshold depression.
[0048] Step 105: Determine whether the user is in a state requiring intervention based on the screening results; if so, extract intervention suggestions from the pre-built subthreshold depression intervention database and generate an intervention strategy based on the intervention suggestions.
[0049] Therefore, the subthreshold depression individualized intervention strategy recommendation method based on multimodal data proposed in this application obtains multimodal psychological data from different populations, conducts subthreshold depression screening training, constructs a screening model, and then generates corresponding subthreshold depression screening results after receiving user data to be identified. Based on the screening results, it determines whether the user is in a state requiring intervention; if so, it extracts intervention suggestions from a pre-built subthreshold depression intervention database and generates intervention strategies based on the intervention suggestions, realizing self-screening and intervention recommendations for users to be identified, and providing accurate and reliable auxiliary recommendation strategies for doctors' diagnosis.
[0050] In some embodiments, Figure 2 An exemplary structural block diagram of a screening and intervention device for subthreshold depression according to an embodiment of this application is shown, which mainly includes: a multimodal data acquisition unit 101, a data preprocessing unit 102, a screening training unit 103, and a data recognition unit 104.
[0051] Among them, the multimodal data acquisition unit 101 is used to acquire multimodal psychological data of normal people, subthreshold depression patients and depression patients.
[0052] A normal person refers to an individual without depressive symptoms. Multimodal data acquisition includes physiological indicators, behavioral performance, cognitive abilities, etc. The multimodal data acquisition unit 101 collects mental health-related data from different angles and methods in research or clinical practice for training the subthreshold depression screening model. In this embodiment, the specific data type of the acquired psychological data is not limited and can be set according to the needs of the application scenario.
[0053] The data preprocessing unit 102 is used to preprocess multimodal psychological data and add it to the database.
[0054] The data preprocessing unit 102 preprocesses the multimodal psychological data with different data sources and types acquired by the multimodal data acquisition unit 101. Preprocessing improves the quality and usability of the data, providing a solid foundation for subsequent comprehensive analysis of the multimodal psychological data. Since the multimodal data acquisition unit 101 does not limit the specific data types acquired, and the data preprocessing methods are set in accordance with the specific data types acquired, the specific data preprocessing methods used in the data preprocessing unit 102 are also not limited here.
[0055] The screening training unit 103 is used to train a pre-built machine learning model for subthreshold depression screening based on a database, and obtain the trained model as a screening model.
[0056] The preprocessed database is divided into training, testing, and validation sets. A pre-built machine learning model is then trained for subthreshold depression screening. In this embodiment, the type and structure of the machine learning model are not limited; random forest models and neural network models can be used. During training, optimization algorithms are used to update the model parameters. After training, the test set is fed into different pre-trained models for validation. Finally, the optimized model is obtained as the subthreshold depression screening model.
[0057] The data recognition unit 104 is used to receive user data to be identified, input it into the screening model for identification, and generate screening results for subthreshold depression.
[0058] The system receives user data to be identified. The data type of the user data corresponds to the multimodal psychological data used in training, and will not be elaborated further here. The received user data is input into the trained screening model. The screening model identifies subthreshold depression based on the user's feature vectors on the emotion network indicator, outputting a probability value. The higher the probability value, the higher the likelihood of subthreshold depression. This probability value can be directly used as the screening result for subthreshold depression. Alternatively, a threshold can be set; if the probability value reaches or exceeds the threshold, it is determined to be subthreshold depression; if it is less than the threshold, it is determined not to be subthreshold depression. Two screening results for subthreshold depression are described here; other methods can be referred to the description in this embodiment, and will not be elaborated further here.
[0059] Based on the above description, the device provided in this embodiment acquires multimodal psychological data and calls a machine learning model to identify subthreshold depression based on the user's feature vectors on the emotion network index. The multimodal psychological data includes multiple data sources such as text, voice, facial expressions, and physiological signals. The fusion of these data can provide more comprehensive information, which helps to more accurately identify the symptoms of subthreshold depression, thereby reducing the number of newly diagnosed patients with depression. At the same time, by using artificial intelligence technology to achieve early screening of subthreshold depression, it can automatically process large amounts of data, quickly generate analysis results, reduce the subjectivity and error of human judgment, and greatly improve the efficiency of screening. It can be easily scaled up to a large population to achieve widespread screening and monitoring.
[0060] This embodiment proposes a specific structure for a multimodal data acquisition unit 101, including: a scale assessment data acquisition subunit, a cognitive test data acquisition subunit, a heart rate variability data acquisition subunit, and a micro-expression video acquisition subunit.
[0061] Scale assessment refers to scales for assessing mental health, such as the General Health Questionnaire (GHQ-12 / GHQ-28), the Patient Health Questionnaire (PHQ-9), the Hamilton Depression Rating Scale (HAM-D), the Beck Depression Rating Scale (BDI), etc. The specific type of scale is not limited here.
[0062] Cognitive test data is data collected through cognitive tests to assess an individual's cognitive functions, such as memory, attention, executive function, language ability, and visuospatial ability. Common tests include the digit span test and the Stroop test.
[0063] Heart rate variability (HRV) refers to the variation in the interval between heartbeats, reflecting the activity state of the autonomic nervous system. Common HRV indicators include standard deviation (SDNN), root mean square of the difference between adjacent RR intervals (RMSSD), low-frequency (LF) power, and high-frequency (HF) power.
[0064] Microexpressions refer to facial muscle changes that occur within a short period of time (usually less than one second). They are often difficult to detect with the naked eye but can reflect true emotional states such as anger, fear, happiness, and sadness. Microexpression video data can provide objective evidence of individual emotional changes and help identify hidden emotional issues.
[0065] Under this unit structure, data such as scale assessments, cognitive tests, heart rate variability, and micro-expressions of normal individuals, subthreshold depression, and depression are collected, uploaded to a cloud server, and a database is established. This multimodal data combination covers multiple aspects such as psychology, cognition, physiology, and behavior, which can more comprehensively reflect the psychological state of an individual and improve the sensitivity and accuracy of early identification.
[0066] In one embodiment, the data preprocessing unit 102 connected to the multimodal data acquisition unit 101 having the above structure may include the following sub-units:
[0067] The numerical data processing subunit is used to extract scale assessment data, cognitive test data, and heart rate variability data as numerical data; to screen the numerical data for missing, abnormal, and logical errors; and to add the screened data to the database.
[0068] The video data processing subunit is used to extract micro-expression videos frame by frame into numerical data as micro-expression data; input the micro-expression data into a classification model for action classification, obtain the output action classification results, and add them to the database.
[0069] The collected data is preprocessed, and numerical data such as scales, cognitive tests, and heart rate variability are checked to handle missing values, outliers, and logical errors. Micro-expression video data can be processed into numerical data frame by frame using the open-source toolkit Py-Feat. The preprocessed micro-expression data is then put into a classification model (such as a random forest model) for classification. The output action classification results are combined with scales, cognitive tests, heart rate variability, etc., to build a database.
[0070] The data preprocessing unit is structured to divide the collected data into numerical data and video data for separate preprocessing. This allows for the adoption of the most suitable processing methods for different types of data, improving data quality and computational efficiency, enhancing model performance, and facilitating data fusion and interpretation. This, in turn, more effectively supports the identification and analysis of subthreshold depression.
[0071] The video data processing subunit may specifically include the following subunits:
[0072] The frame extraction subunit is used to extract micro-expression video frames into numerical data, which is used as micro-expression data.
[0073] The action classification subunit is used to input micro-expression data into the classification model to classify actions and obtain the output action classification results.
[0074] The results evaluation subunit is used to calculate the mean DecreaseGini (MDG) index for each action classification result;
[0075] The data is divided into sub-units to average the Gini index from largest to smallest defects, extract the action classification results corresponding to the top N defects, and add them to the database. Here, N is a positive integer, and its specific value can be set according to actual usage requirements; no limitation is imposed here.
[0076] The mean diminishing Gini index (MDI) is a metric for measuring feature importance, reflecting the contribution of a feature to the purity of nodes within a decision tree. A higher MDI value indicates a stronger discriminative power of the feature when partitioning data. In this video data processing sub-unit structure, selecting the action units with the highest MDI as input variables ensures that the model uses the most discriminative features, thereby improving its predictive performance.
[0077] In one embodiment, the subthreshold depression screening and intervention device may further include an intervention unit connected to the data recognition unit.
[0078] The intervention unit specifically includes:
[0079] The status determination subunit is used to determine whether a user is in a state requiring intervention based on the screening results; if so, it triggers the intervention plan output subunit.
[0080] The intervention plan output subunit is used to extract intervention suggestions from a pre-built subthreshold depression intervention library and generate an intervention plan based on the intervention suggestions.
[0081] The sources of the subthreshold depression intervention database include, but are not limited to: (1) multimodal psychological data from mobile phones of the multimodal data acquisition unit; (2) clinical data: medical diagnosis data and treatment plan data of patients obtained by hospitals or medical research institutions; (3) management and intervention guidelines and expert consensus on subthreshold depression and depression: such expert consensus and guidelines obtained from authoritative academic journals at home and abroad; (4) datasets related to subthreshold depression obtained by web crawlers from medical professional websites, medical institutions and forums. The above data can be processed using Python to initially establish a subthreshold depression professional knowledge dataset, and then after expert screening, the final subthreshold depression intervention professional knowledge dataset can be obtained. The subthreshold depression intervention database can be generated by fine-tuning using Python.
[0082] This embodiment proposes an innovative screening and management model for subthreshold depression. By setting up intervention units, the screening model is combined with individualized intervention to accurately identify subthreshold depression and provide individualized intervention. This allows for more rational allocation of medical resources and reduces unnecessary resource expenditures.
[0083] In one embodiment, the intervention plan output subunit specifically includes:
[0084] The typology determination subunit is used to determine the type of subthreshold depression based on screening results;
[0085] The output subunit is used to extract intervention suggestions corresponding to the type from the pre-built subthreshold depression intervention library and generate an intervention plan based on the intervention suggestions.
[0086] The intervention plan output subunit within this structure identifies different types of subthreshold depression based on each user's specific circumstances, provides intervention suggestions, and generates a personalized intervention plan. Individualized intervention plans are generally better suited to meet the needs and preferences of patients, improving the effectiveness and targeted nature of treatment. The types of subthreshold depression can be symptom types, such as mood symptoms (depressed mood, loss of interest, irritability, or restlessness), cognitive symptoms (difficulty concentrating, negative thinking, decreased self-esteem, etc.), physical symptoms (sleep disturbances, decreased appetite, fatigue, and lack of energy, etc.), behavioral symptoms (social withdrawal, decreased work efficiency, etc.), and psychological symptoms (anxiety, fear, etc.); or the severity of symptoms, such as mild, moderate, or severe, without limitation here.
[0087] Furthermore, in one embodiment, the intervention unit may also include an intervention library optimization subunit. The intervention library optimization subunit is used to acquire user follow-up data after the intervention; and to optimize the intervention recommendations in the subthreshold depression intervention library based on the user follow-up data.
[0088] After individualized intervention for patients with subthreshold depression, regular follow-up and outcome tracking are conducted. Based on the outcome (e.g., turning to normal, turning to depression, or subthreshold depression), the characteristics of different categories are analyzed, which can further improve the individualized intervention method.
[0089] Further reference Figure 3 It illustrates an exemplary structural block diagram of a screening and intervention system for subthreshold depression according to an embodiment of this application, which mainly includes:
[0090] Multimodal data acquisition equipment is used to collect multimodal psychological data from normal individuals, subthreshold depression patients, and patients with depression.
[0091] The device for screening and intervening in subthreshold depression is used to acquire multimodal psychological data, preprocess the multimodal psychological data, and add it to a database; train a pre-built machine learning model for subthreshold depression screening based on the database, and obtain the trained model as the screening model; receive user data to be identified, input it into the screening model for identification, and generate subthreshold depression screening results.
[0092] The specific unit settings and unit working modes in the screening and intervention device for subthreshold depression can be referred to the above embodiments, and will not be repeated here.
[0093] In one embodiment, the multimodal data acquisition device may specifically include:
[0094] Smartphones are used to collect scale assessment data and cognitive test data from normal individuals, patients with subthreshold depression, and patients with depression.
[0095] Depth cameras are used to capture micro-expression videos of normal people, subthreshold depression patients, and patients with depression.
[0096] A dynamic electrocardiogram (ECG) recorder is used to collect heart rate variability data from normal individuals, patients with subthreshold depression, and patients with depression.
[0097] The above is a common combination of devices that can be used to collect scale assessment data, cognitive test data, depth camera data, and dynamic electrocardiogram recording data. Other devices that can achieve the above functions, such as smartwatches, can also be used, and this is not a limitation.
[0098] In one embodiment, the subthreshold depression screening and intervention device is further used to determine whether the user is in a state requiring intervention based on the screening results; if so, it extracts intervention suggestions from a pre-built subthreshold depression intervention database and generates an intervention plan based on the intervention suggestions.
[0099] Accordingly, a smart wearable device can be further set up in the screening and intervention system for subthreshold depression; the smart wearable device is connected to the screening and intervention device for subthreshold depression and is used to output the intervention plan at regular intervals.
[0100] Smart wearable devices are small electronic devices that integrate sensors, processors, and connectivity functions. They are typically worn on the body or embedded in clothing, such as smartwatches and health bracelets. Based on the user's physiological and psychological data, the system can generate personalized intervention plans, including exercise suggestions, sleep guidance, and relaxation exercises. Through smart wearable devices, users can conveniently view their health data and intervention suggestions, which is convenient, fast, and not limited by time or location, enhancing user engagement and motivation. Furthermore, smart wearable devices can monitor and collect various physiological and environmental data, such as heart rate, steps, sleep quality, and blood pressure, and transmit the data wirelessly (e.g., via Bluetooth or Wi-Fi) to smartphones or other terminal devices for analysis and display. In this embodiment, the smart wearable device can also serve as a member of the user-side multimodal data acquisition device, achieving more accurate diagnosis and intervention suggestions through the integration and analysis of multi-source data; however, this is not a limitation.
[0101] The block diagrams in the accompanying drawings illustrate the architecture of possible implementations of apparatuses and systems according to various embodiments of this application. In this regard, each block in the block diagram may represent a module, segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two connected blocks may actually execute substantially in parallel, or they may sometimes execute in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operational instructions, or using a combination of dedicated hardware and computer instructions.
[0102] The units described in the embodiments of this application can be implemented in software or hardware. The described units can also be located in a processor. The names of these units or modules do not, in some cases, constitute a limitation on the unit itself.
[0103] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the foregoing disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.
Claims
1. A method for recommending individualized intervention strategies for subthreshold depression based on multimodal data, characterized in that, include: Acquire multimodal psychological data from normal individuals, patients with subthreshold depression, and patients with depression; The multimodal psychological data is preprocessed and added to the database; The pre-built machine learning model is trained for subthreshold depression screening based on the database, and the trained model is used as the screening model. Receive user data to be identified, input it into the screening model for identification, and generate screening results for subthreshold depression; Based on the screening results, determine whether the user is in a state requiring intervention; If so, intervention suggestions are extracted from a pre-built subthreshold depression intervention database, and intervention strategies are generated based on the intervention suggestions.
2. The method for recommending individualized intervention strategies for subthreshold depression based on multimodal data according to claim 1, characterized in that, The multimodal psychological data includes: scale assessment data, cognitive test data, heart rate variability data, and micro-expression videos.
3. The method for recommending individualized intervention strategies for subthreshold depression based on multimodal data according to claim 1, characterized in that, The preprocessing of the multimodal psychological data and its addition to the database includes: The scale assessment data, the cognitive test data, and the heart rate variability data are extracted as numerical data; the numerical data are screened for missing, abnormal, and logically incorrect data; the screened data is then added to the database. The micro-expression video is extracted frame by frame into numerical data, which is used as micro-expression data; the micro-expression data is input into a classification model for action classification, the output action classification results are obtained, and added to the database.
4. The method for recommending individualized intervention strategies for subthreshold depression based on multimodal data according to claim 3, characterized in that, Specifically, it includes: The micro-expression video is extracted frame by frame into numerical data, which is used as micro-expression data; The micro-expression data is input into a classification model to classify actions, and the output action classification results are obtained. Calculate the average Gini index for each of the action classification results; The top N values of the average reduction in the Gini index from largest to smallest disadvantage are used to extract the action classification results corresponding to the top N values and add them to the database.
5. The method for recommending individualized intervention strategies for subthreshold depression based on multimodal data according to claim 1, characterized in that, The step of extracting intervention suggestions from a pre-built subthreshold depression intervention database and generating intervention strategies based on the intervention suggestions includes: The type of subthreshold depression was determined based on the screening results; Intervention suggestions corresponding to the aforementioned types are extracted from a pre-built subthreshold depression intervention database, and intervention strategies are generated based on these intervention suggestions.
6. A device for recommending individualized intervention strategies for subthreshold depression based on multimodal data, characterized in that, include: The modal data acquisition unit is used to acquire multimodal psychological data from normal individuals, subthreshold depression patients, and depression patients. The data preprocessing unit is used to preprocess the multimodal psychological data and add it to the database; The screening training unit is used to train a pre-built machine learning model for subthreshold depression screening based on the database, and obtain the trained model as a screening model. The data recognition unit receives user data to be identified, inputs it into the screening model for recognition, and generates screening results for subthreshold depression. An intervention recommendation unit is used to determine whether a user is in a state requiring intervention based on the screening results. If so, extract intervention suggestions from the pre-built subthreshold depression intervention database and generate an intervention plan based on the intervention suggestions.
7. The device for recommending individualized intervention strategies for subthreshold depression based on multimodal data according to claim 6, characterized in that, The intervention unit further includes an intervention library optimization subunit, used to acquire user follow-up data after intervention; and to optimize the intervention recommendations in the subthreshold depression intervention library based on the user follow-up data.
8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the method for recommending individualized intervention strategies for subthreshold depression based on multimodal data as described in any one of claims 1-5.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, this program implements the method for recommending individualized intervention strategies for subthreshold depression based on multimodal data as described in any of claims 1-5.
10. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the method for recommending individualized intervention strategies for subthreshold depression based on multimodal data, as described in any one of claims 1-5.