Depression tendency identification and mind opening method and system

Abstract

The invention discloses a depression tendency recognition and mind opening method and system, and the method comprises the steps: constructing a VR virtual interaction scene, the VR virtual interaction scene comprises a cognitive task scene and a free exploration scene, the cognitive task scene is used for recognizing the depression tendency of a user, and the free exploration scene is used for carrying out mind opening on the user; the subject realizes an interaction behavior with the cognitive task scene by means of the VR equipment, and feedback data information of the subject is collected at the same time; according to the feedback data information, the emotional state of the subject is obtained, the emotional state comprises a normal state and a depression state, and if the emotional state is the depression state, the next step is carried out; the subject interacts with the free exploration scene by means of the VR equipment, feedback data information of the subject is collected and compared with the normal state of the subject, and the opening effect of the free exploration scene is evaluated.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence emotion recognition, and in particular to a method and system for identifying depressive tendencies and guiding mindfulness. Background Art

[0002] Virtual reality technology, as a computer simulation system for creating and experiencing a virtual world, provides a multi-sensory, highly interactive, three-dimensional dynamic visual experience by generating a simulated environment, allowing users to deeply experience this environment. This technology has made significant progress in the medical field, especially in the combined application of brain-computer interfaces, lower limb exoskeleton rehabilitation robots and virtual reality. For example, the research team led by neuroscientist Miguel Nicolelis of Duke University in the United States has achieved effective rehabilitation for patients with spinal cord injuries.

[0003] The brain is the most complex and sophisticated organ in the human body, responsible for controlling functions such as movement, sensation, cognition, memory and emotions. The study of brain science is crucial for understanding the cognitive mechanism of the brain, treating brain-related diseases (such as depression), and promoting the development of artificial intelligence and other fields. In addition, the research and development of brain-computer interface technology is also of great significance for exploring the operating mechanism of the brain and promoting human health and scientific progress.

[0004] Functional near-infrared spectroscopy (fNIRS) is an emerging optical imaging technology that uses near-infrared light with a wavelength of 700-900nm to measure blood flow dynamics in the brain. It has been proven to effectively measure the neurobiological function of the brain. Compared with other brain imaging technologies, fNIRS has better temporal and spatial resolution, as well as stronger quietness and tolerance to motion artifacts. It is particularly suitable for infants, children and patients who cannot adapt to traditional imaging technologies. In addition, fNIRS equipment is portable and low-cost, making it suitable for physiological and clinical research in natural social interaction environments.

[0005] With the generation of a large amount of time series data, traditional statistical methods are no longer sufficient to effectively analyze fNIRS data, while the application of deep learning technology can effectively extract features from these data. Therefore, it is particularly important to develop a method for emotion recognition by combining fNIRS and virtual reality technology. Summary of the invention

[0006] In order to overcome the above-mentioned shortcomings and deficiencies of the prior art, the object of the present invention is to provide a method and system for identifying depressive tendencies and guiding mindfulness, and specifically to a method and system for identifying depressive tendencies and guiding mindfulness based on fNIRS and VR scenes.

[0007] This method improves the accuracy of emotion recognition by performing deep learning on the interaction of virtual scenes and the acquisition of fNIRS signals.

[0008] The purpose of the present invention is achieved through the following technical solutions:

[0009] A method for identifying depressive tendencies and guiding them with mindfulness, including:

[0010] Constructing a VR virtual interaction scene, wherein the VR virtual interaction scene includes a cognitive task scene and a free exploration scene, wherein the cognitive task scene is used to identify the user's depressive tendency, and the free exploration scene is used to guide the user to be mindful;

[0011] The subjects interact with the cognitive task scene with the help of VR equipment, and the feedback data information of the subjects is collected at the same time;

[0012] According to the feedback data information, the emotional state of the subject is obtained, and the emotional state includes a normal state and a depressed state. If the emotional state is a depressed state, proceed to the next step;

[0013] The subjects interacted with the free exploration scene with the help of VR equipment. At the same time, the subjects' feedback data information was collected and compared with the subjects' normal state to evaluate the enlightenment effect of the free exploration scene.

[0014] Furthermore, the cognitive task scenarios include an aquarium scenario and an urban virtual reality scenario;

[0015] The aquarium scene includes an aquarium, tables and chairs, walking visitors and noisy background sounds. The aquarium is equipped with two large central rocks, scattered rocks, corals, water plants and swimming fish. After entering the scene, the subject is located in front of the table, facing the aquarium. When looking down, he can see a panel on the table for displaying target fish. The task mode uses the non-X task in CPT. There are two types of stimuli in this scene, including visual stimulation and auditory stimulation, which appear alternately.

[0016] The urban virtual reality scene includes ten identifiable potential target locations, and adds urban environmental sound effects, various types of road sign arrows and voice guidance.

[0017] Furthermore, in the urban virtual reality scene, the subjects first understand the scene task requirements according to the voice prompts, and then begin to follow the predetermined route to learn all potential target locations. During this process, the subjects move along a fixed route. During this stage, the subjects move their perspective to observe the map scene and 10 potential target locations. At the same time, they can press the trigger button on the left handle to control the speed. During this process, the potential target buildings will have yellow light columns to increase the subject's attention and help the subject remember the potential target location. The time required for the preset route depends on the subject himself. Letting the subject control the speed of movement can effectively relieve dizziness, while increasing the subject's participation and interactivity, and ensuring the subject's participation.

[0018] Furthermore, free exploration scenes include gardening scenes and valley scenes.

[0019] Furthermore, the gardening scene includes watering flowers and playing with dogs, and the gardening scene provides a natural environment and an interactive space.

[0020] Furthermore, the valley scene is set as a mountain lake valley with open water and mountains, and the interactive behaviors include fishing and bird feeding;

[0021] The fishing interaction runs when the subject approaches water, and the feeding the birds interaction runs when the subject grabs bird food.

[0022] Furthermore, the behavior of the fishing rod is adjusted through the status obtained by the VR device.

[0023] Furthermore, the feedback data information includes fNIRS data, eye movement data, head movement data and electrodermal data.

[0024] Furthermore, the eye movement data includes the number and proportion of gaze points in the extracted interest area, the total reading time, the first gaze time, whether to accept the return gaze, whether the first main viewpoint is a return gaze, the first gaze time of the second reading, the time of the first passing through the interest area, whether a return gaze is issued from the interest area, whether a return gaze is issued when passing through the interest area for the first time, and whether the interest area is gazed at during the first reading.

[0025] A system based on the above-mentioned method for identifying depression tendency and guiding mindfulness, comprising:

[0026] VR virtual interaction scene module, which is used to build VR virtual interaction scenes including cognitive task scenes and free exploration scenes. The cognitive task scenes are used to identify users' depressive tendencies, and the free exploration scenes are used to guide users in mindfulness.

[0027] Emotion recognition module: The subjects use VR equipment to interact with cognitive task scenarios, and collect feedback data from the subjects;

[0028] According to the feedback data information, the emotional state of the subject is obtained, and the emotional state includes a normal state and a depressed state;

[0029] Emotional guidance module: The subjects interact with the free exploration scene with the help of VR equipment. At the same time, the subjects' feedback data information is collected and compared with the subjects' normal state to evaluate the guidance effect of the free exploration scene.

[0030] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0031] (1) The present invention uses VR scenes to identify and guide emotions, which generates less physiological data with low value density and high accuracy, thus solving the problem that current methods for identifying and guiding depressive tendencies are not accurate enough.

[0032] (2) The present invention constructs cognitive task scenarios and free exploration scenarios simultaneously, realizes interactive behaviors with the help of VR devices, creates an immersive experience environment for users, and serves as a trigger source for emotions and physiological data during the scenario construction stage. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 is a partial schematic diagram of a horticultural planting scene in an embodiment of the present invention;

[0034] Figure 2 is a partial schematic diagram of an aquarium scene in an embodiment of the present invention;

[0035] Figure 3 It is a work flow chart of the present invention. DETAILED DESCRIPTION

[0036] The present invention will be further described in detail below in conjunction with examples, but the embodiments of the present invention are not limited thereto.

[0037] Example 1

[0038] Figure 1-Figure 3 As shown, a method for identifying depressive tendencies and guiding them with mindfulness includes:

[0039] A VR virtual interaction scene is constructed, wherein the VR virtual interaction scene includes a cognitive task scene and a free exploration scene. The cognitive task scene is used to identify the user's depressive tendency, and the free exploration scene is used to guide the user in mindfulness.

[0040] The subjects interacted with the cognitive task scenarios with the help of VR equipment, and their feedback data was collected at the same time.

[0041] According to the feedback data information, the emotional state of the subject is obtained, and the emotional state includes a normal state and a depressed state. If the emotional state is a depressed state, proceed to the next step;

[0042] The subjects interacted with the free exploration scene with the help of VR equipment. At the same time, the subjects' feedback data information was collected and compared with the subjects' normal state to evaluate the enlightenment effect of the free exploration scene.

[0043] The cognitive task scenarios include an aquarium scenario and an urban virtual reality scenario.

[0044] Furthermore, the aquarium scene consists of an aquarium, a set of tables and chairs, walking visitors and noisy background sounds. There are two large central rocks, scattered small rocks, corals, water plants and swimming small fish in the aquarium. After entering the scene, the subject is located in front of the table, facing the aquarium, and can see a panel on the table for displaying special fish when looking down. The task mode uses the non-X task in CPT (Continuous Performance Test), that is, the participants need to respond to stimuli other than the target stimulus. There are two types of stimuli - visual stimulation and auditory stimulation. The two stimuli appear randomly, and there is a 6-second interval between each stimulus. After the test starts, the randomly generated fish will appear from behind the large rock on the left, swim to the large rock on the right and disappear. In addition, the name of the fish will be randomly broadcast in the headphones, and the subject needs to press the button when he sees or hears fish other than special fish.

[0045] Test process: After entering the scene, the subject will see a line of text in front of him that says "Please pull the trigger of the left handle to start the tutorial". After pulling the left trigger button, the subject will enter the teaching mode. At the beginning of the tutorial, the subject will see the prompt "Please press the disc on the right handle when you see or hear fish other than specific fish". After 5 seconds, the prompt disappears and the task begins. During the task, if the subject responds correctly, a prompt "Correct" will appear in front of him. If the response is wrong, "Wrong" will be displayed. If there is an omission, "Miss" will be displayed. The tutorial lasts 1 minute. After the tutorial, the subject will see the instruction "Please pull the trigger of the right handle to start the test". After pulling the trigger button of the right handle, the subject will enter the formal test. There will be no text prompts during the formal test, and the test lasts 3 minutes.

[0046] Furthermore, the urban virtual reality scene is used to detect the user's depression level through an urban virtual scene, including 10 identifiable potential target locations. In addition, common urban environmental sound effects are added to the virtual scene accordingly to enhance the user's immersion. In addition, the system has built-in multiple types of road sign arrows and voice guidance, which can be placed in the scene to help users better remember their routes.

[0047] The implementation process of this scenario is as follows:

[0048] It consists of three parts: preset routes, random building target selection, and target search scripts.

[0049] The preset route script controls the user's three-dimensional coordinate changes to achieve the movement effect. 17 points are created in the urban scene, and the script is written to make the user move at a constant speed between points. When reaching the point in front of the potential target building, it pauses for 7 seconds to allow the subject to remember the geographical location of the potential target point; it pauses for 0.5 seconds at the turning position of the intersection to simulate the actual turning effect. The starting point and the end point of the final preset route are the same, and the preset route goes around the external road of the map, covering all potential target building points.

[0050] When the preset route is completed, the countdown starts for 1 minute. This process is the subject's free exploration time. After 1 minute, the script controls the random building target selection and target search scripts to start executing.

[0051] The building target selection script then labels the potential target building objects as `Protential Targets`. After execution begins, 3 of the 10 Potential Targets are randomly selected as target buildings. The target building label changes to Finding Targets. At the same time, the corresponding miniature versions of these target buildings are selected and labeled TipTargets.

[0052] In the target search script, the subject is first teleported to a random point. A total of 15 random teleportation points are set in the map, which are evenly distributed in the map scene and are all on open roads. After the teleportation is completed, the FindingTargets label object is found, and the TipTargets label object is teleported to the subject's child object point. Its physical coordinates are below the subject's perspective, and the user only needs to lower his head slightly to view the target building style. At the same time, the detection function runs. For the subject's eye movement data, its gaze ray is monitored. If it intersects with the object, it is determined whether the object's label is Finding Targets. If it meets the requirements, the target judgment part function is started. When the distance to the target building is less than or equal to 15 in-game distance units and the ray intersects with the target building for 0.5s, it is determined that the object has been found.

[0053] At the beginning or end of each stage or when the subject finds the target building, the `indexReporter` script automatically sends a tag to the fNIRS device to record the fNIRS signal changes around that timestamp.

[0054] Existing Technology In VR environment, how does ray detection reflect the user's sight and intention? Figure 1 This is a difficult point. The VR scene of the present invention adopts the following means:

[0055] Optimization of hierarchy and physics collision bodies: I optimized the hierarchy and physics collision bodies of ray detection, only detecting key interactive elements, thereby reducing unnecessary performance overhead.

[0056] Using Unity’s built-in physics system: By leveraging Unity’s Raycast feature, I was able to implement raycasting in an efficient manner, which was crucial to keeping the app running smoothly.

[0057] Event-driven interaction: I manage raycasting through events triggered based on user behavior and game logic. This ensures that the raycasting logic is both simple and effective.

[0058] The free exploration scenes include a gardening scene and a valley scene.

[0059] Furthermore, the valley scene includes setting up a mountain lake valley with open water and mountains, and the interactive behaviors include fishing and bird feeding; when the subject approaches the water, the fishing interaction is run; when the subject grabs bird food, the bird feeding interaction is run.

[0060] Specifically: In the VR scene, when approaching water, you can run an interactive script for fishing. After waiting for a while, the fishing rod will shake and the handle will vibrate to remind the user that the fish has been hooked. The user can increase the fishing progress by pressing and holding the touchpad of the VR controller. I use the intensity of the handle vibration to indicate the progress of fishing, and the intensity of the handle vibration will decrease as the progress increases. When the fish is successfully caught, the fish is hung on the pole. The user can move the fish to the fish net in the water by moving the handle, and the captured fish will swim in the fish net, enhancing the positive feedback of fishing interaction through intuitive visual stimulation.

[0061] When the user grabs the bird food, the bird feeding interaction will start. Each bird has a timer. When the timer value is greater than 0, the bird will follow the randomly moving target point. When the timer value is less than 0, the bird will follow the character and move around the character. When the distance between the thrown bird food and the bird is less than a certain threshold, the bird will fly to the bird food to complete the feeding action and reset the timer.

[0062] In the interactive fishing action, the state obtained by the VR device is used to adjust the behavior of the fishing rod, specifically adjusting the subject's seat or the position of the fishing hook.

[0063] The gardening planting scene includes watering flowers and playing with dogs, and the gardening planting scene provides a natural environment and an interactive space.

[0064] In the gardening planting scene in this embodiment, the subject will enter a garden surrounded by a sea of ​​flowers, where he can walk freely in the garden and the sea of ​​flowers, water the vegetables and enjoy the pastoral scenery. The garden is similar to a back garden in reality, with a plastic greenhouse, and a variety of vegetables and crops are planted indoors and outdoors: some grow directly on the ground, some grow in the seedbed, some grow with the help of a bracket, and some grow in a pot... The subject can pick up a watering can and water any crop in the garden. After successful watering, the plants will grow, for example, the buds will grow into flowers, and mature crops will bear fruit. The subject can also pick the fruits of the crops with ripe fruits and put them in a basket. There is also a cute puppy in the garden, which can walk freely and interact with the patient. There are many interactive styles, and the dog with multiple skills is committed to making the subject feel accompanied and happy. The sea of ​​flowers is full of flowers of various types and shapes, which will sway in the wind and are randomly distributed. The sea of ​​flowers has no end, it is an infinite space, and the range of activities of the subject will not be restricted, avoiding the negative effects that may be caused by locking the subject in a limited space. The scenes contain natural and interactive sounds, which make the scenes more realistic, enhance the patient's sense of immersion, and assist in relaxation.

[0065] This VR-based restorative environment provides a natural environment and interactive space, which can promote parasympathetic nerve activity, make subjects feel safe and relaxed, effectively relieve physical and mental stress, and improve mental health. With the help of an eye tracker to judge the subject's line of sight, the interactive activities are simple and unified, easy to learn, and can provide timely feedback, ensuring that they will not burden patients.

[0066] Furthermore, the feedback data information in the present invention includes fNIRS data, eye movement data, head movement data and electrodermal data.

[0067] Furthermore, the feedback data information also includes a pre-processing step, specifically:

[0068] The data preprocessing is intended to perform preliminary processing on the collected fNIRS, eye movement, and electrodermal data.

[0069] The fNIRS data preprocessing refers to converting the collected raw light intensity data into optical density first, and then converting it into hemoglobin concentration data through the modified Beer-Lambert law, eliminating useless noise signals through filtering, and removing motion artifacts through baseline calibration.

[0070] The eye movement data preprocessing refers to extracting 11 indicators, including the number and proportion of fixation points in the interest area, the total reading time, the first fixation time, whether to accept the return gaze, whether the first main viewpoint is the return gaze, the first fixation time of the second reading, the time of the first passing through the interest area, whether the return gaze is issued from the interest area, whether the return gaze is issued when passing through the interest area for the first time, and whether the interest area is fixed during the first reading, and generating corresponding sorting files for users to extract other indicators by themselves.

[0071] The electrodermal data preprocessing is to first filter out data of poor quality, select a time window value as the baseline value, a time window value as the maximum value of the response, subtract the two to obtain the original value induced by the stimulus, and finally perform data correction.

[0072] Furthermore, based on the feedback data information, the emotional state of the subject is obtained. Specifically, the VR video data of the stimulus source and the fNIRS, eye movement, and electrocutaneous physiological data of the source subject's response are used to train a symptom recognition model to infer whether the target subject has a tendency to depression. If the emotional state is judged to be a depressive state, the positive guidance step is entered.

[0073] The training symptom recognition model includes a VR video guidance module, a physiological response module, and a depression recognition module.

[0074] The VR video guidance module first divides the video sequence into multiple time periods for processing, and extracts the features of the original video frame and the dynamic feature map respectively through two feature extractors in each time period, and finally splices the feature information of different time periods.

[0075] The physiological response module refers to extracting the features of three physiological signals respectively through fnris, eye movement and skin electrical data encoder, and fusing the features by splicing method.

[0076] The depression recognition module fuses the feature vectors output by the VR video guidance module and the physiological response module through an attention mechanism, and finally outputs a prediction result by a multi-layer perceptron to determine whether the subject has a tendency to depression.

[0077] Furthermore, the subjects interacted with the free exploration scene with the help of VR equipment, and the feedback data of the subjects was collected and compared with the normal state of the subjects to evaluate the enlightenment effect of the free exploration scene, specifically:

[0078] Using the fnirs, eye movement, and electrocutaneous physiological data of the source subject, the corresponding data feature vectors are first extracted through three encoders, and then the three modal data are aligned and mapped to a potential code based on the MoPoE multimodal coding block, which can effectively record the information of the three data. In the inference stage, the recovery of the subject's condition is determined by evaluating the similarity between the encoding obtained by the target subject's physiological data input and the encoding in the depression or normal state obtained in the training stage.

[0079] The encoder is a multi-layer perceptron with two hidden layers.

[0080] The MoPoE-based multimodal encoding block is a multimodal automatic encoding variational Bayesian model, which mainly encodes the features of three types of data: fnirs, eye movements, and electrodermal activity.

[0081] The evaluation of the enlightenment effect of the free exploration scenario is specifically conducted in the form of a questionnaire.

[0082] This embodiment also provides a system for identifying depression tendency and guiding mindfulness, including:

[0083] VR virtual interaction scene module, which is used to build VR virtual interaction scenes including cognitive task scenes and free exploration scenes. The cognitive task scenes are used to identify users' depressive tendencies, and the free exploration scenes are used to guide users in mindfulness.

[0084] Emotion recognition module: also includes a data collection module, where the subjects use VR equipment to interact with the cognitive task scene and collect feedback data information from the subjects;

[0085] According to the feedback data information, the emotional state of the subject is obtained, and the emotional state includes a normal state and a depressed state;

[0086] Emotional guidance module: The subjects interact with the free exploration scene with the help of VR equipment. At the same time, the subjects' feedback data information is collected and compared with the subjects' normal state to evaluate the guidance effect of the free exploration scene.

[0087] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be equivalent replacement methods and are included in the protection scope of the present invention.

Claims

1. A method for identifying depressive tendencies and guiding mindfulness, characterized in that: include: Constructing a VR virtual interaction scene, wherein the VR virtual interaction scene includes a cognitive task scene and a free exploration scene, wherein the cognitive task scene is used to identify the user's depressive tendency, and the free exploration scene is used to guide the user to be mindful; The subjects interact with the cognitive task scene with the help of VR equipment, and the feedback data information of the subjects is collected at the same time; According to the feedback data information, the emotional state of the subject is obtained, and the emotional state includes a normal state and a depressed state. If the emotional state is a depressed state, proceed to the next step; The subjects interacted with the free exploration scene with the help of VR equipment. At the same time, the subjects' feedback data information was collected and compared with the subjects' normal state to evaluate the enlightenment effect of the free exploration scene.

2. The method for identifying depression tendency and guiding mindfulness according to claim 1, characterized in that: The cognitive task scenarios include an aquarium scenario and an urban virtual reality scenario; The aquarium scene includes an aquarium, tables and chairs, walking visitors and noisy background sounds. The aquarium is equipped with two large central rocks, scattered rocks, corals, water plants and swimming fish. After entering the scene, the subject is located in front of the table, facing the aquarium. When looking down, he can see a panel on the table for displaying target fish. The task mode uses the non-X task in CPT. There are two types of stimuli in this scene, including visual stimulation and auditory stimulation, which appear alternately. The urban virtual reality scene includes ten identifiable potential target locations, and adds urban environmental sound effects, various types of road sign arrows and voice guidance.

3. The method for identifying depression tendency and guiding mindfulness according to claim 2, characterized in that: In the urban virtual reality scene, the subjects first understand the scene task requirements according to the voice prompts, and then begin to follow the predetermined route to learn all potential target locations. During this process, the subjects move along a fixed route. During this stage, the subjects move their perspective to observe the map scene and 10 potential target locations. At the same time, they can press the trigger button on the left handle to control the speed. During this process, there will be yellow light beams on the potential target buildings to increase the subjects' attention and help the subjects remember the potential target locations. The time required for the preset route depends on the subjects themselves. Letting the subjects control the movement speed by themselves can effectively relieve dizziness, while increasing the subjects' participation and interactivity, and ensuring the subjects' participation.

4. The method for identifying depression tendency and guiding mindfulness according to claim 1, characterized in that: Free exploration scenes include a gardening scene and a valley scene.

5. The method for identifying depression tendency and guiding mindfulness according to claim 4, characterized in that: The gardening planting scene includes watering flowers and playing with dogs, and the gardening planting scene provides a natural environment and an interactive space.

6. The method for identifying depression tendency and guiding mindfulness according to claim 4, characterized in that: The valley scene is set in a mountain lake valley with open water and mountains, and interactive behaviors include fishing and bird feeding; The fishing interaction runs when the subject approaches water, and the feeding the birds interaction runs when the subject grabs bird food.

7. The method for identifying depression tendency and guiding mindfulness according to claim 6, characterized in that: Adjust the behavior of the fishing rod based on the state obtained from the VR device.

8. The method for identifying depression tendency and guiding mindfulness according to claim 1, characterized in that: The feedback data information includes fNIRS data, eye movement data, head movement data and electrodermal data.

9. The method for identifying depression tendency and guiding mindfulness according to claim 8, characterized in that: The eye movement data includes the number and proportion of gaze points in the extracted interest area, the total reading time, the first gaze time, whether to accept the return gaze, whether the first main viewpoint is a return gaze, the first gaze time of the second reading, the time of the first passing through the interest area, whether a return gaze is issued from the interest area, whether a return gaze is issued when passing through the interest area for the first time, and whether the interest area is gazed at during the first reading.

10. A system based on the method for identifying depression tendency and guiding mindfulness according to any one of claims 1 to 9, characterized in that: include: VR virtual interaction scene module, which is used to build VR virtual interaction scenes including cognitive task scenes and free exploration scenes. The cognitive task scenes are used to identify users' depressive tendencies, and the free exploration scenes are used to guide users in mindfulness. Emotion recognition module: The subjects use VR equipment to interact with cognitive task scenarios, and collect feedback data from the subjects; According to the feedback data information, the emotional state of the subject is obtained, and the emotional state includes a normal state and a depressed state; Emotional guidance module: The subjects interact with the free exploration scene with the help of VR equipment. At the same time, the subjects' feedback data information is collected and compared with the subjects' normal state to evaluate the guidance effect of the free exploration scene.

Images