Multi-modal dementia risk reduction intervention system for promoting social participation

Abstract

The invention discloses a multi-modal dementia risk reduction intervention system for promoting social participation, and belongs to the technical field of dementia intervention, and the system specifically comprises a user behavior collection module which is used for collecting and correcting a basic behavior set of a user; the behavior symbol conversion module is used for converting the basic behavior set into a behavior coding sequence formed by standardized behavior symbols; the intervention path generation module is used for generating an intervention instruction set from the behavior coding sequence, extracting corresponding interaction materials from the multi-modal content library and generating collaborative intervention content; the real-time behavior monitoring module is used for pushing collaborative intervention content to the user and monitoring real-time behavior feedback of the user; the behavior expectation comparison module is used for carrying out difference degree calculation on the real-time behavior feedback and preset behavior expectation, and when the difference degree is larger than a preset threshold value, the intervention instruction set is triggered to be reconfigured; according to the method, the precision and the dynamics of social participation and promotion of the senile dementia patients are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of dementia intervention, and in particular to a multimodal dementia risk reduction intervention system for promoting social participation. Background Art

[0002] As the global population ages at an accelerated pace, dementia prevention has become a public health priority. Epidemiological studies have confirmed that sustained social participation can effectively maintain cognitive reserve and reduce the risk of neurodegenerative diseases. The World Health Organization has clearly stated that promoting social connections and multi-dimensional activity participation among middle-aged and elderly people is a core strategy for dementia risk intervention. Current mainstream medical opinion emphasizes the need to build a comprehensive intervention program that integrates physical activities, social interactions, and cognitive stimulation to achieve multi-target synergistic activation of neural pathways.

[0003] In the existing technology, computer-assisted dementia risk intervention programs mainly develop in two directions: one type adopts a standardized cognitive training paradigm, providing structured training through preset memory games or logic task software; the other type adopts a behavioral monitoring trigger mechanism, pushing general recommendations when sensors detect specific physiological indicators.

[0004] However, standardized cognitive training tasks are disconnected from real social scenarios, making it difficult to reflect the behavioral interaction patterns of individuals in daily environments, and unable to make real-time adjustments based on individual behavioral patterns. For example, forcing socially avoidant people to engage in group activities may cause resistance. Intervention dimensions such as physical activities, social interactions, and cognitive stimulation lack organic integration. For example, simple VR social training is not combined with real-world movement tasks, resulting in low user participation. In addition, the evaluation of intervention effects mostly relies on regular clinical scales, and it is impossible to dynamically reconstruct the intervention path based on behavioral feedback, resulting in delayed strategy adjustments.

[0005] These limitations make it difficult for existing systems to adapt to the spatiotemporal variability of individual behavior patterns, and it is also impossible to achieve coordinated optimization of multimodal intervention elements. Summary of the Invention

[0006] The purpose of the present invention is to provide a multimodal dementia risk reduction intervention system that promotes social participation and solves the following technical problems: Existing dementia intervention systems have difficulty adapting to the temporal and spatial variability of individual behavioral patterns.

[0007] The purpose of the present invention can be achieved through the following technical solutions: A multimodal dementia risk reduction intervention system that promotes social engagement, including: User behavior collection module, used to collect and modify the user's basic behavior set, including the user's displacement records in physical space, interaction frequency records in social scenarios, and multimedia contact content records; The behavior symbol conversion module is used to convert the basic behavior set into a behavior coding sequence composed of standardized behavior symbols through a preset standardized symbol library. Each behavior symbol corresponds to a set social participation behavior pattern. The intervention path generation module is used to generate an intervention instruction set containing a symbol replacement sequence and a symbol insertion sequence from a behavior coding sequence. According to the intervention instruction set, the corresponding interactive materials are extracted from the multimodal content library to generate collaborative intervention content with spatial movement guidance, social task guidance, and media content guidance. A real-time behavior monitoring module is used to push collaborative intervention content to users through portable terminals and monitor real-time behavioral feedback during user execution; The behavior expectation comparison module is used to calculate the difference between the real-time behavior feedback and the behavior expectation preset in the intervention instruction set. When the difference is greater than the preset threshold, the intervention path generation module is triggered to reconfigure the intervention instruction set.

[0008] As a further solution of the present invention: In the user behavior file, the process of modifying the user's basic behavior set is: A meteorological data interface is integrated into the displacement record, and when the rainfall intensity exceeds the rainfall intensity threshold, the moving distance threshold corresponding to the distance symbol is automatically lowered; Introducing acoustic environment analysis into the interaction frequency record, and reducing the allocation priority of dialogue symbols when the background noise decibel exceeds the noise tolerance threshold; Integrating light sensor data into the multimedia contact content record to shield the request for generating text symbols when the ambient illumination is lower than the reading illumination standard; The timestamp overlap ratio of the above three records is calculated. If the timestamp overlap ratio is higher than the overlap ratio threshold, the corrected behavior set is input into the behavior symbol conversion module; otherwise, the timestamp is verified.

[0009] As a further solution of the present invention: in the behavior symbol conversion module, the standardized symbol library specifically includes: The standardized symbol library includes a spatial displacement symbol set, a social interaction symbol set, and a media contact symbol set; the spatial displacement symbol set includes distance symbols representing different movement distance thresholds, stay symbols representing different stay time thresholds, and speed symbols representing different movement speed thresholds; The social interaction symbol set includes conversation symbols for face-to-face conversation, collaboration symbols for group activities, and stranger symbols for stranger contact; The media engagement symbol set includes text symbols for reading text, auditory symbols for listening to audio, and visual symbols for watching images.

[0010] As a further solution of the present invention: in the behavior symbol conversion module, the process of converting the basic behavior set into a behavior coding sequence composed of standardized behavior symbols is as follows: Obtain the user's displacement record in the physical space. When the continuous movement distance exceeds the preset distance threshold, a corresponding distance symbol is assigned. When the length of stay at a specific location reaches the stay time threshold, a stay symbol is assigned. When the movement speed is lower than the speed threshold, a low speed symbol is assigned. Event recognition is performed on the interaction frequency records in user social scenarios. When a two-way language communication event is recognized, a conversation symbol is assigned. When a joint activity of three or more people is detected, a collaboration symbol is assigned. When contact with an object whose identity information is not stored is captured, a stranger symbol is assigned. Determine the media type of the user's multimedia contact content records, assign a text symbol when the contact content is written text, assign an auditory symbol when it is pure audio information, and assign a visual symbol when it is a dynamic image; Arrange all assigned symbols in chronological order to form a behavioral coding sequence with timestamp marks.

[0011] As a further solution of the present invention: in the intervention path generation module, the generation process of the intervention instruction set is: Obtain a symbol association rule library and a symbol priority weight table. The symbol association rule library stores the spatial proximity, social complementarity, and media compatibility relationships between behavioral symbols. The symbol priority weight table records the quantitative value of the intensity of neural function stimulation of each type of symbol, which is divided into high weight, medium weight, and low weight. Scan the behavioral coding sequence and mark low-frequency symbols, where the low-frequency symbols are symbols that appear less than the average number of symbols of the same type within a set time period. Search the symbol association rule library for target symbols that have a strong association with the low-frequency symbols, where the strong association is defined as a spatial proximity score, a social complementarity score, and a media compatibility score all exceeding corresponding thresholds. Generate a symbol insertion sequence containing the target symbol, where the insertion position is determined based on the temporal gap density in the behavioral coding sequence. Identify high-risk symbol combinations in behavior coding sequences, where the high-risk symbol combinations refer to symbol segments in which the number of consecutively appearing symbols of the same type exceeds a type continuity threshold and the total priority weight is lower than a weight threshold; generate a symbol replacement sequence, and replace a set proportion of original symbols in the high-risk symbol combination with high-weight symbols, with the replacement position preferentially selecting the symbol position in the middle of the time sequence in the high-risk symbol combination.

[0012] As a further solution of the present invention: the mechanism for identifying high-risk symbol combinations in the behavior coding sequence is: Record the user's brainwave characteristic change amplitude and the next day's memory recall test score after each symbol combination execution; when a new symbol combination satisfies the conditions where both the brainwave characteristic change amplitude and the memory recall test score are lower than the corresponding benchmark values, the symbol combination will be included in the high-risk symbol combination library.

[0013] As a further solution of the present invention: in the intervention path generation module, the generation process of the collaborative intervention content is: Spatial movement guidance consists of a geofence coordinate set and a movement path animation. The geofence coordinate set generates a circular activity area based on the movement distance threshold corresponding to the distance symbol in the symbol insertion sequence. The movement path animation draws a recommended route to the target geofence starting from the current location. Social task guidance adopts a dual-track system of role-playing dialogue scripts and collaborative task lists. The role-playing dialogue script generates a virtual dialogue tree containing open-ended questions based on dialogue symbols, while the collaborative task list is designed based on collaborative symbols to carry out physical object transfer tasks that require the cooperation of multiple people. Media content guides the implementation of a cross-media mapping strategy. When the target symbol is a text symbol, a news summary with pictures is generated; when the target symbol is an auditory symbol, an audio clip with subtitles is generated; when the target symbol is a visual symbol, an interactive dynamic infographic is generated; All collaborative intervention contents are arranged on the timeline according to the symbolic order of the intervention instruction set. There are overlapping periods between spatial movement guidance and social task guidance, while media content guidance runs through the entire intervention cycle.

[0014] As a further solution of the present invention: in the real-time behavior monitoring module, the process of pushing collaborative intervention content to the user via the portable terminal is as follows: Spatial movement guidance is based on augmented reality technology. It superimposes geo-fence boundary lines and movement direction arrows on the real-time image of the terminal camera, triggering a visual confirmation signal when the user enters the geo-fence area. Social tasks guide the implementation of double verification. Conversation tasks use a microphone to collect voice responses and perform keyword matching verification. Collaboration tasks complete contact verification by scanning the QR code generated by the other user's terminal. Media content guidance is based on user gaze detection. When the front camera detects that the user's gaze deviates from the screen by more than the gaze deviation angle, content playback is paused until the user's gaze refocuses on the screen and reaches the continuous gaze duration, then playback continues.

[0015] As a further solution of the present invention: In the behavior expectation comparison module, the specific process of difference calculation and reconfiguration is as follows: Extract symbolic response delay data and symbolic execution completeness data from real-time behavioral feedback. The symbolic response delay data records the time difference from pushing the collaborative intervention content to detecting the corresponding behavioral symbol. The symbolic execution completeness data records the feature matching degree between the actual generated symbol and the expected symbol. When the symbol response delay exceeds the delay tolerance threshold, delay compensation reconfiguration is triggered to add a transition symbol before the symbol insertion sequence, wherein the transition symbol has a strong correlation with the target symbol and has a lower priority weight than the target symbol; When the symbol execution completeness is lower than the completeness qualification threshold, complexity degradation reconfiguration is triggered to replace high-weight symbols in the symbol replacement sequence with medium-weight symbols of the same type, while reducing the number of target symbols in the symbol insertion sequence; The reconfigured intervention instruction set retains the total time length of the original instruction set and keeps the total length unchanged by extending a single symbolic execution period or adding parallel symbolic execution channels.

[0016] Beneficial effects of the present invention: (1) This invention establishes a digital archive of individual daily behavior trajectories and implements behavioral symbolization conversion, converting three types of raw data, namely physical displacement, social contact, and media interaction, into standardized behavioral symbol sequences with social participation attributes, thereby achieving digital mapping of complex behavioral patterns in real social scenarios and eliminating the semantic gap between standardized cognitive training tasks and real social activities.

[0017] (2) The present invention relies on a dynamic intervention path generation unit to construct a symbol association rule library and a priority weight table. Based on the identification of low-frequency symbols and high-risk symbol combinations, it generates a symbol insertion sequence and a symbol replacement sequence. Based on this, it extracts three-dimensional elements of spatial movement guidance, social task guidance, and media content guidance from the multimodal content library. Through the time axis coordination of the spatial positioning of the geo-fence coordinate set and the movement path animation, the social drive of the role dialogue script and the collaborative task list, and the cognitive activation of the cross-media mapping strategy, it realizes the organic integration of physical activities, social interactions, and cognitive stimulation in a real environment, and overcomes the participation defect of virtual training being disconnected from real movement.

[0018] (3) Based on real-time behavioral feedback data monitored by portable terminals, the present invention triggers the reconfiguration of the intervention instruction set by calculating the difference between symbol response delay and symbol execution completeness. The environmental parameter correction module integrates meteorological, acoustic, and optical data to dynamically adjust the symbol allocation logic, forming a closed-loop response mechanism to replace the traditional periodic clinical scale assessment model, ensuring minute-by-minute dynamic optimization of intervention strategies in response to individual behavioral patterns and environmental changes. Through environmentally adaptive data processing, multi-dimensional collaborative intervention, and a closed-loop feedback mechanism, the present invention achieves precise and dynamic promotion of social participation in patients with Alzheimer's disease. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The present invention will be further described below with reference to the accompanying drawings.

[0020] Figure 1 It is a module schematic diagram of the present invention. DETAILED DESCRIPTION

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0022] See also Figure 1 As shown, the present invention is a multimodal dementia risk reduction intervention system for promoting social participation, comprising: The user behavior collection module uses sensors to obtain physical displacement, social interaction and multimedia contact data, and corrects the data based on environmental parameters such as meteorology, acoustics, and lighting. For example, it lowers the movement distance threshold during rainfall, reduces the priority of dialogue symbols when noise exceeds the standard, and blocks text symbol generation when light is insufficient. It also ensures data consistency through timestamp overlap verification.

[0023] The behavioral symbol conversion module is based on a standardized symbol library. It converts displacement records into spatial symbols according to distance, duration of stay, and speed; converts social interactions into social symbols according to conversation, collaboration, and stranger contact; and converts multimedia contacts into media symbols according to text, audio, and image. It generates a behavioral coding sequence with timestamps in chronological order.

[0024] The intervention path generation module, based on a symbol association rule library and a priority weight table, marks low-frequency symbols and retrieves strongly associated target symbols to generate insertion sequences. It also identifies high-risk combinations of consecutive symbols of the same type that exceed a threshold and have insufficient total weights to generate replacement sequences. Based on this, it extracts material from a multimodal content library and generates collaborative intervention content combining geofence navigation, social task scripts, and cross-media content. Spatial movement and social task guidance are delivered simultaneously during overlapping periods, with media content integrated throughout the entire process.

[0025] The real-time behavior monitoring module pushes intervention content via portable devices. Spatial movement guidance uses AR to overlay fence boundaries and directional arrows. Social tasks are verified through voice keyword matching and QR code scanning, and media content is paused based on gaze detection. The behavior expectation comparison module calculates the difference between response latency and execution completeness. When thresholds are exceeded, reconfiguration is triggered, such as inserting transition symbols to compensate for delays and replacing weighted symbols to reduce complexity. Dynamic optimization is achieved by adjusting the execution period to maintain the intervention duration constant.

[0026] In a preferred embodiment of the present invention, the process of modifying the user's basic behavior set in the user behavior profile is as follows: For physical space displacement records, the system obtains real-time rainfall intensity information through a meteorological data interface. When rainfall intensity exceeds a preset threshold, it automatically adjusts the distance symbol's corresponding movement distance judgment criteria. For example, while 500 meters is typically used as the distance threshold for moderate-intensity activity, this threshold can be dynamically lowered to 200 meters during rainfall, preventing misjudgments of user mobility due to weather factors.

[0027] When recording social interaction frequency, the system incorporates an acoustic environment analysis module. This module uses the device's microphone to collect real-time background noise decibel levels. When the noise level exceeds the threshold for comfortable communication, the system automatically lowers the priority of assigning conversational symbols. This mechanism ensures that brief and ineffective voice interactions in high-noise environments like square dancing are not mistakenly identified as valid social interactions, thus improving data validity.

[0028] To record multimedia content, the system integrates data from the terminal device's light sensor and continuously monitors ambient illumination levels. When the ambient illumination is detected to be below the standard reading illumination, the system automatically blocks requests for text symbol generation, preventing low-light text browsing from being recorded as effective cognitive stimulation. At the same time, the system prioritizes the generation of auditory or visual symbols.

[0029] After completing the environmental correction for these three types of data, the overlap ratio of timestamps for movement, social interaction, and multimedia engagement records is further calculated. By comparing the consistency of the time tags of these three types of data, it is determined whether they represent collaborative behaviors occurring at the same time. If the overlap ratio exceeds a preset threshold, it indicates that the data has spatiotemporal correlation and can be input into the behavioral symbol conversion module. If the overlap ratio is insufficient, the timestamp verification process is initiated. Through mechanisms such as sensor clock synchronization calibration and data transmission delay compensation, the time deviation is corrected to ensure the accuracy of subsequent behavioral analysis.

[0030] In another preferred embodiment of the present invention, in the behavior symbol conversion module, the standardized symbol library specifically includes: The semantic mapping of user behavior is achieved through a standardized symbol library, which uses a three-tiered classification system to construct a behavioral semantic network. The library's spatial displacement symbol set is based on the spatial attributes of physical activities. Distance symbols are divided into multiple levels of granularity based on movement distance thresholds. For example, a continuous movement of 500 meters is set as the basic distance symbol, while 1000 meters is set as the high-intensity distance symbol. Different symbols are used to identify differences in user activity intensity. Stay symbols are determined based on the duration of stay at a specific location. For example, a stay of more than 30 minutes at a community activity center is assigned a long-stay symbol, while a stay of less than 10 minutes is assigned a short-stay symbol. This distinguishes between passive and active social stops. Speed ​​symbols include low, medium, and high speeds. A low-speed symbol is assigned when the user's walking speed is less than 1.5 km / h, while a high-speed symbol is assigned when the user's walking speed is greater than 5 km / h. This allows for a more refined description of movement status.

[0031] The social interaction symbol set focuses on the scene characteristics and relationship attributes of interpersonal interaction. The allocation of dialogue symbols must meet the conditions of two-way language communication. The system uses voice recognition technology to detect turn-taking in the dialogue. When both parties speak at least twice in a single communication, it is judged to be a valid dialogue and symbols are allocated; collaboration symbols are aimed at common activity scenarios of three or more people, such as community painting groups, square dance teams, etc., and video analysis technology is used to identify group collaborative behavior patterns. Symbol allocation is triggered when task division or action synchronization is detected among members; stranger symbols are used to mark contact behaviors with objects whose identity information is not stored. The system uses face recognition technology to compare the local identity database. When the interaction object is not in the preset contact list, a stranger symbol is allocated regardless of the length of the communication to quantify the social participation of non-acquaintances.

[0032] The media contact symbol set is divided according to the medium characteristics of the information carrier. Text symbols cover all types of written text contact behaviors, including paper books, electronic documents, community bulletin board texts, etc. The system identifies pure text content through file type detection and content analysis technology; auditory symbols correspond to pure audio information contact scenarios, such as radio programs, voice chats, audiobooks, etc., and exclude multimedia content containing videos through audio spectrum analysis; visual symbols are aimed at dynamic image carriers, including movies, short videos, animations, etc., and are judged by detecting frame sequence changes and visual element motion characteristics in the content. Static image contact is not included in the symbol category.

[0033] In a preferred embodiment of the present invention, in the behavior symbol conversion module, the process of converting the basic behavior set into a behavior coding sequence composed of standardized behavior symbols is as follows: The symbol allocation for displacement records adopts a threshold trigger mechanism. The system calculates the cumulative moving distance of the GPS trajectory in real time. When the continuous moving distance exceeds the 500-meter threshold, the corresponding distance symbol is automatically allocated, and the stay time timer is started at the same time. If the stay at a certain geographic coordinate point exceeds 15 minutes, a stay symbol is allocated; the speed judgment is based on the displacement change per unit time. The instantaneous speed is calculated after smoothing the positioning data through the Kalman filter algorithm. When it is lower than the threshold, a low-speed symbol is allocated.

[0034] Event recognition of social interaction records integrates audio energy detection and visual posture analysis. Two-way language communication events must simultaneously meet the characteristics of alternating changes in voice signal energy and interactive facial orientation. Group activities of more than three people use human posture estimation technology to count the number of human key points in the same scene, and trigger collaborative symbol allocation when the threshold is exceeded. Strange contact recognition combines facial recognition confidence and interaction distance sensor data, and assigns strange symbols when unfamiliar objects enter the social distance of 1.5 meters.

[0035] A multi-level detection strategy is used to determine the media type of multimedia contact records. First, the file extension is used to preliminarily distinguish text, audio, and video formats. Then, an in-depth analysis of the content is performed. Text files must contain recognizable text character sequences. Audio files must exclude video tracks and their spectra must conform to speech or music characteristics. The dynamic images corresponding to visual symbols must have a frame rate greater than 12 per second and pixel displacement changes.

[0036] The behavior represented by each symbol is related to participation in social activities. For example, the distance symbol corresponds to social participation in going out and contacting the external environment; the dialogue symbol corresponds to social participation in interpersonal communication and emotional interaction; and the collaboration symbol corresponds to social participation in group cooperation and establishing social relationships.

[0037] All assigned symbols are sorted by the timestamp of the event, with a granularity of seconds. This forms a temporal behavioral coding sequence, such as "[T1: distance symbol 500 meters], [T2: conversation symbol], [T3: visual symbol short video]," providing standardized behavioral semantic input for subsequent intervention path generation. This symbol conversion mechanism uses multi-dimensional feature detection and thresholding to map raw behavioral data into semantic units with social engagement attributes, achieving an abstract upgrade from physical signals to behavioral patterns.

[0038] In another preferred embodiment of the present invention, in the intervention path generation module, the generation process of the intervention instruction set is: The intervention path generation module constructs a decision-making core using a symbol association rule library and a priority weight table to intelligently generate intervention instruction sets. The symbol association rule library uses a triple structure to store the relationships between behavioral symbols, including spatial proximity, social complementarity, and media compatibility. Spatial proximity describes the probability of co-occurrence of symbols corresponding to behaviors in physical space. For example, if the co-occurrence probability of the "500-meter distance symbol" and the "dialogue symbol" in a community park exceeds 70%, it is considered strong spatial proximity. Social complementarity quantifies the synergistic effect of different social symbols on social engagement. For example, the combination of "collaboration symbols" and "unfamiliarity symbols" can increase social network expansion efficiency by 40%. Media compatibility assesses the complementarity of different media symbols in information transmission. For example, the combination of "text symbols" and "auditory symbols" can increase information memory retention by 30%. The symbol priority weight table, based on neuroscience research, quantifies the stimulation intensity of symbols on brain regions such as the prefrontal cortex and hippocampus into three levels: high, medium, and low. "Collaboration symbols" are given high weights due to the coordinated activation of multiple brain regions, while "stay symbols" are given low weights due to their low stimulation intensity.

[0039] The module first scans the behavioral coding sequence for low-frequency symbols. Using a sliding time window, it calculates the frequency of each symbol within a set period, such as seven days, and labels symbols that are less than 60% of the average for symbols of the same type as low-frequency symbols. For example, in the social interaction symbol set, if a "strange symbol" appears less than twice per week, meaning less than the average of five times for its type, the association retrieval mechanism is triggered. In the symbol association rule base, strongly associated target symbols must simultaneously meet three criteria: a spatial proximity score of at least 0.7, calculated based on the probability of co-occurrence between the two symbols in historical data; a social complementarity score of at least 0.6, determined based on an evaluation of social network expansion effectiveness; and a media compatibility score of at least 0.5, determined through information transmission efficiency testing. Retrieved target symbols, such as the "collaboration symbol," are added to the symbol insertion sequence. Insertion locations are determined using a temporal gap density algorithm: This algorithm identifies periods of time within the behavioral sequence where no symbols are recorded for more than 15 minutes. New symbols are prioritized during high-density gaps, such as the regular idle hours between 2 and 4 p.m. on weekdays, to avoid disrupting the user's natural rhythm.

[0040] To identify high-risk symbol combinations, the module employs a dual-threshold detection mechanism: first, it determines whether there are consecutive occurrences of the same type of symbols, such as three or more consecutive "stay symbols" within a spatial displacement symbol cluster. Secondly, it calculates the total priority weight of the symbol segment. If it falls below the weight threshold, for example, if the total value is less than 1.5 due to the continuous appearance of low-weight symbols, the combination is marked as high-risk. When generating symbol replacement sequences, the "center-first" principle is followed: for high-risk segments containing five consecutive "stay symbols," the third symbol, the "collaboration symbol" with a high weight in the middle of the chronological order, is replaced first, thereby increasing the overall weight of the symbol segment with the minimum intervention cost. The replacement ratio is dynamically adjusted based on the length of the symbol segment, with a 20%-30% replacement for segments of 5-7 symbols and a 30%-40% replacement for segments of 8 or more symbols, ensuring that the intervention intensity matches the risk level.

[0041] In a preferred embodiment of the present invention, the mechanism for identifying high-risk symbol combinations in the behavior coding sequence is: Using EEG data acquisition equipment such as a dry electrode EEG headband, the user's alpha and beta wave amplitude changes after executing a symbol combination are recorded. Simultaneously, a mobile app is used to conduct a next-day memory recall test, including episodic memory tasks such as recalling the clothing characteristics of yesterday's social contacts and semantic memory tasks such as retelling key information from a text encountered yesterday. When the EEG characteristic change of a new symbol combination is 30% lower than the historical average and the memory test score is lower than the baseline, such as a score of less than 6 out of 10, the combination is automatically added to the high-risk symbol combination library, triggering a rule library update process. For example, if the consecutive occurrence of "text symbol" combinations results in a 20% decrease in hippocampal activation and a text content recall accuracy rate of less than 50%, it will be marked as high-risk. The weight threshold for subsequent similar combinations will be automatically lowered by 15% to improve detection sensitivity.

[0042] In another preferred embodiment of the present invention, in the intervention path generation module, the generation process of the collaborative intervention content is: The spatial movement guidance system uses geofencing technology to construct physical activity scenarios. Based on the distance threshold corresponding to the distance symbols in the symbol insertion sequence, the system generates a circular activity area centered on the user's current location on an electronic map. For example, a 500-meter movement threshold generates a circular area with a radius of 250 meters. The movement path animation module uses real-time positioning as a starting point and combines street view data with pedestrian path distribution to draw a recommended route to the target geofence. This route prioritizes frequent social gatherings such as community parks and convenience stores, converting abstract distance symbols into actionable physical movement instructions.

[0043] Social task guidance adopts a dual-track design to adapt to the needs of different social scenarios. The character dialogue script module generates a virtual dialogue tree based on dialogue symbols, containing open-ended questions such as "What changes have you seen in the community's greening recently?" Users can interact through voice or text input, and the depth of the dialogue tree dynamically adjusts based on the user's historical social data. The collaborative task list module designs physical object delivery tasks based on collaborative symbols, such as "Give the community activity center brochure to the volunteer in the blue shirt." The task includes the characteristics of the target object, a description of the delivered object, and the completion location. Users must cooperate with others to complete the task, and the system detects collaborative contact status through near-field communication or Bluetooth signals.

[0044] Media content guides the implementation of a cross-modal mapping strategy to achieve multi-sensory information delivery. When the target symbol is a text symbol, the system generates a news summary with 3-5 accompanying images, with each article limited to 200 words. When the target symbol is an auditory symbol, a 5-10 minute audio clip with subtitles is generated, with the subtitle font size dynamically adjusted based on ambient lighting. When the target symbol is a visual symbol, an interactive dynamic infographic is generated, allowing users to click on a hotspot to view detailed information, such as clicking on a venue icon on a community event map to display opening hours and other information.

[0045] In terms of timeline arrangement, all collaborative intervention elements were arranged in the symbolic order of the intervention instruction set. Spatial movement guidance and social task guidance overlapped within the geofenced area. For example, upon arriving at the community plaza, collaborative tasks within the plaza were automatically triggered, achieving a scenario-coupling of physical movement and social interaction. Media content guidance was implemented throughout the intervention cycle, with auditory content delivered during walking sessions and visual infographics displayed during social breaks to create continuous cognitive stimulation.

[0046] In a preferred embodiment of the present invention, in the real-time behavior monitoring module, the process of pushing collaborative intervention content to the user via the portable terminal is as follows: Spatial movement guidance is based on augmented reality technology. A translucent geographic fence boundary line and an arrow-shaped movement indicator are superimposed on the real-time image of the terminal camera. When the user enters the fenced area, a flashing green light effect is triggered at the edge of the screen as a visual confirmation signal. The duration of the light effect is associated with the preset duration of the stay symbol.

[0047] Social tasks guide the implementation of a two-factor authentication mechanism to ensure interaction validity. For conversational tasks, voice responses are captured through a microphone and keywords are extracted using natural language processing technology. Interactions are considered valid if they match at least 60% of the pre-set keyword library in the conversation script. Collaborative tasks require users to scan a dynamic QR code generated by the other party's terminal. The QR code contains the task ID and the identity of the collaborative partner. After a successful scan, the system verifies that the interaction was completed within the designated area using GPS positioning.

[0048] Media content guidance integrates eye tracking technology to optimize information reception efficiency. The terminal's front-facing camera uses an iris detection algorithm to monitor gaze direction. If the user's gaze deviates from the screen by more than 15 degrees for three seconds, content playback is automatically paused, and a semi-transparent notification box is displayed on the screen. When the user's gaze refocuses on the screen for more than two seconds, content playback resumes from the pause point, improving the effective reach of media content.

[0049] In another preferred embodiment of the present invention, in the behavior expectation comparison module, the specific process of difference calculation and reconfiguration is: First, two types of key data are extracted from real-time behavioral feedback: symbol response delay data records the time difference from the moment the collaborative intervention content is pushed to the moment the corresponding behavioral symbol is detected. For example, after the community activity task corresponding to the "collaboration symbol" is pushed, the time interval of the user's actual participation in the activity is recorded; symbol execution completeness data uses a feature matching algorithm to evaluate the similarity between the actual generated symbol and the expected symbol. For example, for a reading task expected to be a "text symbol", the system will analyze whether the actual contact content contains text features and whether the reading time meets the requirements.

[0050] When the symbol response delay exceeds the preset delay tolerance threshold, the system triggers a delay compensation reconfiguration mechanism. This mechanism adds transition symbols before the symbol insertion sequence. The selection of transition symbols follows two principles: a strong correlation with the target symbol to ensure the consistency of behavioral guidance; and a lower priority weight than the target symbol to avoid overstimulating the user. For example, when a user delays responding to a social task corresponding to an "unfamiliar symbol," the system inserts a simple greeting task corresponding to a "conversation symbol" before the original task, such as "asking the community security guard about today's weather." These transition tasks belong to the same social interaction symbol set as the target task, but have lower cognitive complexity, helping users gradually enter the state.

[0051] If the symbol execution completeness falls below the completeness qualification threshold, the system initiates complexity downgrade reconfiguration. Specific measures include replacing high-weight symbols in the symbol replacement sequence with medium-weight symbols of the same type, while reducing the number of target symbols in the symbol insertion sequence. For example, the original plan was to replace consecutive "stay symbols" with group activities corresponding to "collaboration symbols." However, when the user's execution completeness is insufficient, the system will downgrade the group activities to two-person cooperative tasks, such as "organizing the community bookshelf with neighbors," and reduce the frequency of inserting similar tasks. This adjustment not only maintains the consistency of the intervention direction, but also reduces the difficulty of execution, which is in line with the user's current ability status.

[0052] The reconfigured intervention instruction set strictly preserves the total duration of the original instruction set, achieving time balance through two strategies. When the execution period of a single symbol is extended, the system intelligently compresses the execution time of other symbols. For example, the reading task of "text symbols" is extended from 15 minutes to 20 minutes, while the subsequent viewing task of "visual symbols" is shortened from 10 minutes to 5 minutes. If parallel symbol execution channels are added, the system can superimpose low-complexity tasks within the same time window. For example, while a user is listening to a broadcast of "auditory symbols," the simple hand movement task corresponding to the "stay symbol" is simultaneously executed, maximizing time utilization.

[0053] The above is a detailed description of an embodiment of the present invention. However, the content described is only a preferred embodiment of the present invention and should not be considered to limit the scope of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A multimodal dementia risk reduction intervention system that promotes social participation, characterized in that: include: User behavior collection module, used to collect and modify the user's basic behavior set, including the user's displacement records in physical space, interaction frequency records in social scenarios, and multimedia contact content records; The behavior symbol conversion module is used to convert the basic behavior set into a behavior coding sequence composed of standardized behavior symbols through a preset standardized symbol library. Each behavior symbol corresponds to a set social participation behavior pattern. The intervention path generation module is used to generate an intervention instruction set containing a symbol replacement sequence and a symbol insertion sequence from a behavior coding sequence. According to the intervention instruction set, the corresponding interactive materials are extracted from the multimodal content library to generate collaborative intervention content with spatial movement guidance, social task guidance, and media content guidance. A real-time behavior monitoring module is used to push collaborative intervention content to users through portable terminals and monitor real-time behavioral feedback during user execution; The behavior expectation comparison module is used to calculate the difference between the real-time behavior feedback and the behavior expectation preset in the intervention instruction set. When the difference is greater than the preset threshold, the intervention path generation module is triggered to reconfigure the intervention instruction set.

2. A multimodal dementia risk reduction intervention system for promoting social participation according to claim 1, characterized in that: In the user behavior file, the process of modifying the user's basic behavior set is as follows: A meteorological data interface is integrated into the displacement record, and when the rainfall intensity exceeds the rainfall intensity threshold, the moving distance threshold corresponding to the distance symbol is automatically lowered; Introducing acoustic environment analysis into the interaction frequency record, and reducing the allocation priority of dialogue symbols when the background noise decibel exceeds the noise tolerance threshold; Integrating light sensor data into the multimedia contact content record to shield the request for generating text symbols when the ambient illumination is lower than the reading illumination standard; The timestamp overlap ratio of the above three records is calculated. If the timestamp overlap ratio is higher than the overlap ratio threshold, the corrected behavior set is input into the behavior symbol conversion module; otherwise, the timestamp is verified.

3. The multimodal dementia risk reduction intervention system for promoting social participation according to claim 1, characterized in that: In the behavior symbol conversion module, the standardized symbol library specifically includes: The standardized symbol library includes a spatial displacement symbol set, a social interaction symbol set, and a media contact symbol set; the spatial displacement symbol set includes distance symbols representing different movement distance thresholds, stay symbols representing different stay time thresholds, and speed symbols representing different movement speed thresholds; The social interaction symbol set includes conversation symbols for face-to-face conversation, collaboration symbols for group activities, and stranger symbols for stranger contact; The media engagement symbol set includes text symbols for text reading, auditory symbols for audio listening, and visual symbols for image viewing.

4. The multimodal dementia risk reduction intervention system for promoting social participation according to claim 3, characterized in that: In the behavior symbol conversion module, the process of converting the basic behavior set into a behavior coding sequence composed of standardized behavior symbols is as follows: Obtain the user's displacement record in the physical space. When the continuous movement distance exceeds the preset distance threshold, a corresponding distance symbol is assigned. When the length of stay at a specific location reaches the stay time threshold, a stay symbol is assigned. When the movement speed is lower than the speed threshold, a low speed symbol is assigned. Event recognition is performed on the interaction frequency records in user social scenarios. When a two-way language communication event is recognized, a conversation symbol is assigned. When a joint activity of three or more people is detected, a collaboration symbol is assigned. When contact with an object whose identity information is not stored is captured, a stranger symbol is assigned. Determine the media type of the user's multimedia contact content records, assign a text symbol when the contact content is written text, assign an auditory symbol when it is pure audio information, and assign a visual symbol when it is a dynamic image; Arrange all assigned symbols in chronological order to form a behavioral coding sequence with timestamp marks.

5. The multimodal dementia risk reduction intervention system for promoting social participation according to claim 1, characterized in that: In the intervention path generation module, the generation process of the intervention instruction set is as follows: Obtain a symbol association rule library and a symbol priority weight table. The symbol association rule library stores the spatial proximity, social complementarity, and media compatibility relationships between behavioral symbols. The symbol priority weight table records the quantitative value of the intensity of neural function stimulation of each type of symbol, which is divided into high weight, medium weight, and low weight. Scan the behavioral coding sequence and mark low-frequency symbols, where the low-frequency symbols are symbols that appear less than the average number of symbols of the same type within a set time period. Search the symbol association rule library for target symbols that have a strong association with the low-frequency symbols, where the strong association is defined as a spatial proximity score, a social complementarity score, and a media compatibility score all exceeding corresponding thresholds. Generate a symbol insertion sequence containing the target symbol, where the insertion position is determined based on the temporal gap density in the behavioral coding sequence. Identifying high-risk symbol combinations in a behavior coding sequence, wherein the high-risk symbol combination refers to a symbol segment in which the number of consecutive symbols of the same type exceeds a type continuity threshold and the total priority weight value is lower than a weight threshold; Generate a symbol replacement sequence, replace the original symbols in the high-risk symbol combination with high-weight symbols in a set proportion, and give priority to the symbol position in the middle of the time sequence in the high-risk symbol combination.

6. The multimodal dementia risk reduction intervention system for promoting social participation according to claim 5, characterized in that: The mechanism for identifying high-risk symbol combinations in behavioral coding sequences is: Record the user's brainwave characteristic change amplitude and the next day's memory recall test score after each symbol combination execution; when a new symbol combination satisfies the conditions where both the brainwave characteristic change amplitude and the memory recall test score are lower than the corresponding benchmark values, the symbol combination will be included in the high-risk symbol combination library.

7. The multimodal dementia risk reduction intervention system for promoting social participation according to claim 1, characterized in that: In the intervention path generation module, the generation process of the collaborative intervention content is as follows: Spatial movement guidance consists of a geofence coordinate set and a movement path animation. The geofence coordinate set generates a circular activity area based on the movement distance threshold corresponding to the distance symbol in the symbol insertion sequence. The movement path animation draws a recommended route to the target geofence starting from the current location. Social task guidance adopts a dual-track system of role-playing dialogue scripts and collaborative task lists. The role-playing dialogue script generates a virtual dialogue tree containing open-ended questions based on dialogue symbols, while the collaborative task list is designed based on collaborative symbols to carry out physical object transfer tasks that require the cooperation of multiple people. Media content guides the implementation of a cross-media mapping strategy. When the target symbol is a text symbol, a news summary with pictures is generated; when the target symbol is an auditory symbol, an audio clip with subtitles is generated; when the target symbol is a visual symbol, an interactive dynamic infographic is generated; All collaborative intervention contents are arranged on the timeline according to the symbolic order of the intervention instruction set. There are overlapping periods between spatial movement guidance and social task guidance, while media content guidance runs through the entire intervention cycle.

8. The multimodal dementia risk reduction intervention system for promoting social participation according to claim 7, characterized in that: In the real-time behavior monitoring module, the process of pushing collaborative intervention content to the user via the portable terminal is as follows: Spatial movement guidance is based on augmented reality technology. It superimposes geo-fence boundary lines and movement direction arrows on the real-time image of the terminal camera, triggering a visual confirmation signal when the user enters the geo-fence area. Social tasks guide the implementation of double verification. Conversation tasks use a microphone to collect voice responses and perform keyword matching verification. Collaboration tasks complete contact verification by scanning the QR code generated by the other user's terminal. Media content guidance is based on user gaze detection. When the front camera detects that the user's gaze deviates from the screen by more than the gaze deviation angle, content playback is paused until the user's gaze refocuses on the screen and reaches the continuous gaze duration, then playback continues.

9. The multimodal dementia risk reduction intervention system for promoting social participation according to claim 5, characterized in that: In the behavior expectation comparison module, the specific process of difference calculation and reconfiguration is as follows: Extract symbolic response delay data and symbolic execution completeness data from real-time behavioral feedback. The symbolic response delay data records the time difference from pushing the collaborative intervention content to detecting the corresponding behavioral symbol. The symbolic execution completeness data records the feature matching degree between the actual generated symbol and the expected symbol. When the symbol response delay exceeds the delay tolerance threshold, delay compensation reconfiguration is triggered to add a transition symbol before the symbol insertion sequence, wherein the transition symbol has a strong correlation with the target symbol and has a lower priority weight than the target symbol; When the symbol execution completeness is lower than the completeness qualification threshold, complexity degradation reconfiguration is triggered to replace high-weight symbols in the symbol replacement sequence with medium-weight symbols of the same type, while reducing the number of target symbols in the symbol insertion sequence; The reconfigured intervention instruction set retains the total time length of the original instruction set and keeps the total length unchanged by extending a single symbolic execution period or adding parallel symbolic execution channels.

Images