Intelligent terminal consciousness disorder interactive awakening system and method based on traditional Chinese medicine rules

Abstract

This invention discloses an intelligent terminal interactive awakening system and method for patients with impaired consciousness based on Traditional Chinese Medicine (TCM) rules, belonging to the field of intelligent medical technology. The system includes a data acquisition module that acquires multidimensional clinical data of patients with impaired consciousness. Through data cleaning and standardization, an initial patient clinical information matrix is ​​constructed to obtain a structured dataset of impaired consciousness features. An intervention plan generation module generates a preliminary intervention intensity curve based on the structured dataset. For this preliminary intervention intensity curve, boundary condition constraint technology is used to set upper and lower limits for intervention intensity and time. Simultaneously, a smoothness weight allocation method is used to adjust the importance parameters of time points to obtain a smooth intervention execution plan. This intelligent terminal interactive awakening system and method for patients with impaired consciousness based on TCM rules achieves precise and personalized intervention for these patients, improves treatment outcomes, and provides new ideas and methods for clinical practice.

Description

Technical Field

[0001] This invention relates to the field of intelligent medical technology, specifically to an intelligent terminal interactive awakening system and method for consciousness disorders based on traditional Chinese medicine principles. Background Technology

[0002] Patient health management and personalized treatment are core areas of modern medical development, playing a crucial role in improving treatment outcomes and optimizing the allocation of medical resources. Especially in the management of chronic and complex diseases, accurately predicting patient needs and developing personalized treatment plans can not only improve patient experience but also reduce medical risks.

[0003] However, existing methods, when processing patient data, often rely too heavily on single statistical models or generalized treatment templates, neglecting the dynamic evolution of individual patient differences and the multidimensional interactive needs during treatment. This makes it difficult for treatment plans to adapt to the actual situation of patients at different stages, especially in long-term treatment, where patient responses may fluctuate due to various factors, and existing methods struggle to effectively capture these changes. Against this backdrop, the core challenge of this research is how to accurately characterize individual differences and predict the dynamic changes in treatment interaction needs within complex and ever-changing patient data. First, factors such as patient age, gender, disease duration, and level of consciousness are intertwined, collectively influencing treatment needs, but existing technologies struggle to integrate these factors into a dynamic, continuous predictive model. Due to the lack of systematic analysis of patient constitution characteristics, especially personalized feature modeling combined with traditional Chinese medicine constitution identification theory, treatment plans lack specificity and cannot generate treatment schedules and stimulus intensity curves adapted to specific patient stages. These two technical factors are closely related: insufficient dynamic modeling of individual differences limits the effective integration of constitution characteristics, while the lack of constitution characteristics exacerbates the limitations of predictive models. Summary of the Invention

[0004] The purpose of this invention is to provide an interactive awakening system and method for mental disorders based on traditional Chinese medicine rules. By integrating multidimensional patient data and combining it with the theory of constitution identification in traditional Chinese medicine, a dynamic prediction model is constructed to generate a personalized treatment schedule and stimulation intensity curve, and to predict possible treatment responses in advance.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an intelligent terminal interactive awakening system for patients with impaired consciousness based on traditional Chinese medicine rules. The system includes a data acquisition module that acquires multidimensional clinical data of patients with impaired consciousness, and through data cleaning and standardization, constructs an initial patient clinical information matrix to obtain a structured dataset of characteristics of impaired consciousness. An intervention plan generation module generates a preliminary intervention intensity curve based on the structured dataset of characteristics of impaired consciousness. For the preliminary intervention intensity curve, boundary condition constraint technology is used to set upper and lower limits for intervention intensity and time. Simultaneously, a smoothness weight allocation method is used to adjust the importance parameters of time points to obtain a smooth intervention execution plan. The intervention plan includes a data feedback and analysis module, which collects real-time feedback data on the improvement of patients' consciousness impairment during the intervention period according to the smooth intervention execution plan. If the feedback data deviates from the preset threshold range, the rule base parameters are adjusted through dynamic rule update technology. At the same time, combined with interpolation error correction and curve continuity verification, the intervention schedule and stimulus intensity curve are updated to determine the optimized intervention execution plan. The personalized adaptation module uses personalized threshold setting technology to adjust the triggering conditions of consciousness arousal intervention according to individual differences for the optimized intervention execution plan. At the same time, the latest feedback data is integrated into the intervention intensity curve through real-time data fusion method to obtain the final dynamic intervention execution logic.

[0006] Preferably, the preliminary intervention intensity curve obtained for the structured consciousness disorder feature dataset includes, for the structured consciousness disorder feature dataset, using pre-established symptom matching rules, combined with patient state classification methods, classifying the patient state into acute phase and remission phase, and determining the patient's consciousness disorder classification result.

[0007] Preferably, the preliminary intervention intensity curve obtained for the structured dataset of consciousness disorder features also includes extracting individualized features of consciousness disorder symptom frequency and duration from the consciousness disorder classification results, using intervention priority ranking logic to determine the patient's core symptom intervention needs, and obtaining a personalized symptom priority vector.

[0008] Preferably, obtaining the preliminary intervention intensity curve for the structured dataset of consciousness disorder features further includes constructing a preliminary intervention time point distribution logic based on personalized symptom priority vectors and historical data weight analysis, and determining the initial consciousness disorder intervention time schedule through a time schedule generation algorithm.

[0009] Preferably, the preliminary intervention intensity curve obtained for the structured consciousness disorder feature dataset also includes dividing the initial consciousness disorder intervention schedule into smaller time units using time interval subdivision logic, and identifying the peak period of consciousness disorder as the core intervention point through key node extraction technology, thus obtaining a detailed intervention time framework.

[0010] Preferably, obtaining the preliminary intervention intensity curve for the structured dataset of consciousness disorder features further includes calculating the initial stimulus intensity from the refined intervention time frame, combined with the intensity decision model, and calculating the stimulus intensity at the intermediate point between adjacent key nodes using the intensity value interpolation formula to obtain the preliminary intervention intensity curve.

[0011] Preferably, the specific formula for adjusting the triggering conditions for intervention in consciousness disorders based on individual differences using personalized threshold setting technology is as follows: ;in, The threshold representing individual i, This represents the average of the historical data of individual i. This represents the standard deviation of the historical data for individual i. This represents the standard deviation adjustment parameter, and i represents the individual number index.

[0012] Preferably, the standard deviation adjustment parameter The specific formula is as follows:

[0013] ;

[0014] in, This represents the standard deviation adjustment parameter. This represents the variance of the feedback data for individual i. This represents the time delay from the start of intervention to the generation of a response in individual i. This represents the frequency at which individual i receives intervention per unit of time, where i represents the individual's index.

[0015] Preferably, the time delay from the start of intervention to the generation of a response in individual i is... Specifically, it is the difference between the time point when individual i first exhibits a measurable response and the time point when individual i first receives intervention.

[0016] The method for interactive awakening of consciousness disorders based on traditional Chinese medicine (TCM) rules in smart terminals employs the aforementioned TCM-based interactive awakening system for smart terminals. The method includes: acquiring multidimensional clinical data of patients with consciousness disorders; constructing an initial patient clinical information matrix through data cleaning and standardization to obtain a structured dataset of consciousness disorder characteristics; obtaining a preliminary intervention intensity curve for the structured dataset; using boundary condition constraint technology to set upper and lower limits for intervention intensity and time, and adjusting the importance parameters of time points through a smoothness weight allocation method to obtain a smooth intervention execution plan; collecting real-time feedback data on the improvement of consciousness disorders during the intervention period based on the smooth intervention execution plan; adjusting rule base parameters through dynamic rule update technology if the feedback data deviates from a preset threshold range, and updating the intervention schedule and stimulus intensity curve by combining interpolation error correction and curve continuity verification to determine an optimized intervention execution plan; and using personalized threshold setting technology to adjust the triggering conditions for consciousness awakening intervention according to individual differences, and integrating the latest feedback data into the intervention intensity curve through real-time data fusion to obtain the final dynamic intervention execution logic.

[0017] As can be seen from the above technical solution, the present invention has the following beneficial effects:

[0018] This intelligent terminal-based interactive awakening system and method for patients with consciousness disorders, based on Traditional Chinese Medicine (TCM) rules, constructs a dataset of consciousness disorder characteristics by acquiring multidimensional clinical data from patients and classifies the disorders. Based on the classification results, individualized features are extracted to determine core symptom intervention needs and generate an initial intervention schedule. Time interval subdivision and key node extraction techniques are employed to identify peak periods of consciousness disorders and refine the intervention timeframe. An intensity decision model is used to calculate stimulus intensity, generating a smooth intervention execution plan. Real-time patient feedback data is collected, rule base parameters are dynamically adjusted, and the intervention plan is updated. Finally, through personalized threshold settings and real-time data fusion, a dynamic intervention execution logic is obtained. This invention achieves precise and personalized intervention for patients with consciousness disorders, effectively alleviating symptoms and improving treatment outcomes, providing new ideas and methods for clinical practice. Attached Figure Description

[0019] Figure 1 This is a system module connection diagram of the intelligent terminal consciousness disorder interactive awakening system based on traditional Chinese medicine rules of the present invention;

[0020] Figure 2 This is a flowchart of the intelligent terminal interactive awakening method for consciousness disorders based on traditional Chinese medicine rules, as described in this invention. Detailed Implementation

[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 creative efforts are within the scope of protection of the present invention.

[0022] like Figure 1 As shown, this invention provides a technical solution: an intelligent terminal interactive awakening system for patients with consciousness disorders based on traditional Chinese medicine rules. The system includes a data acquisition module that acquires multidimensional clinical data of patients with consciousness disorders, and constructs an initial patient clinical information matrix through data cleaning and standardization to obtain a structured dataset of characteristics of consciousness disorders. An intervention plan generation module generates a preliminary intervention intensity curve for the structured dataset of characteristics of consciousness disorders. For the preliminary intervention intensity curve, boundary condition constraint technology is used to set upper and lower limits for intervention intensity and time. At the same time, the importance parameters of time points are adjusted through a smoothness weight allocation method to obtain a smooth intervention execution plan. The data feedback and analysis module collects real-time feedback data on the improvement of patients' consciousness impairment during the intervention, based on a smooth intervention execution plan. If the feedback data deviates from the preset threshold range, the rule base parameters are adjusted through dynamic rule update technology. At the same time, combined with interpolation error correction and curve continuity verification, the intervention schedule and stimulus intensity curve are updated to determine the optimized intervention execution plan. The personalized adaptation module uses personalized threshold setting technology to adjust the triggering conditions for consciousness arousal intervention according to individual differences for the optimized intervention execution plan. At the same time, the latest feedback data is integrated into the intervention intensity curve through real-time data fusion methods to obtain the final dynamic intervention execution logic.

[0023] The main function of the data acquisition module is to acquire multidimensional clinical data of patients with disorders of consciousness and convert this data into a structured dataset of characteristics of disorders of consciousness through a standard processing procedure, so that the system can subsequently execute precise intervention and control logic.

[0024] The first step is the raw data collection and source definition: the collection dimensions cover five categories: physiological indicators, behavioral characteristics, symptom assessment, emotional state, and environmental factors. Each type of data is provided by different devices and systems: physiological indicators include blood pressure, heart rate, body temperature, respiratory rate, EEG signals, etc., with a sampling frequency of once per minute; behavioral indicators include sleep duration and dietary records, with a sampling frequency of once per day; symptom assessment uses a questionnaire-based self-assessment, including the degree of consciousness impairment, duration, and triggers, with an assessment frequency of twice per day; environmental factors such as light, noise, and air pressure are obtained through mobile applications calling local or network APIs, updated once per hour.

[0025] The second step, data cleaning rules, is as follows: Missing value imputation: If an indicator is missing for a certain period, the system will use the mean of the patient's five most recent historical records to fill the gap. For example, if a patient's heart rate data is missing at minute 120, the system will backtrack to the heart rates at minutes 115, 116, 117, 118, and 119, calculate their mean, and use it to fill the gap at minute 120. Outlier removal: Using the distribution range of the patient's historical data as a reference, the system defines outliers as observations exceeding 5% above or below the historical data mean. For example, if the patient's historical heart rate mean is 75, any heart rate record less than 71 or greater than 78 is considered outlier and removed. Duplicate data merging: When multiple similar data points are collected within one minute, the median is used as the representative value for retention. Inconsistency correction: For data with inconsistent units (such as millimeters of mercury versus kilopascals), the system will uniformly convert to the international standard unit millimeters of mercury, with 1 kilopascal equal to 7.5 millimeters of mercury.

[0026] The third step details the data standardization method: All data items are uniformly normalized, with the range limited to 0 to 1. Range standardization is used, which involves subtracting the minimum value from the current value and then dividing by the difference between the maximum and minimum values. The maximum and minimum values ​​are selected based on the patient's historical records for the past 7 days. For example, if a patient's maximum heart rate over the past 7 days is 98 and the minimum is 62, and the current heart rate is 80, then the normalized value is (80 - 62) divided by (98 - 62), resulting in 0.5.

[0027] The fourth step is to construct a clinical information matrix: The standardized data is structured into a matrix along the time dimension, with the horizontal axis representing time (in minutes) and the vertical axis representing the standardized feature values. Based on recording 1440 time points per day (once per minute), each patient has a daily matrix of 1440 rows and several columns, with the number of columns matching the number of features, typically between 30 and 50 columns. Each row represents a snapshot of the patient's overall health status at that time point.

[0028] Step 5, Explanation of Structured Feature Extraction Method: To capture trend signals before and after the onset of consciousness disorders, the system uses a sliding time window technique to segment the data sequence. The default width of the sliding window is 180 minutes. Within a continuous 180 minutes, the system calculates: the difference between the maximum and minimum values, as an indicator of fluctuation intensity; the rate of change of the mean value of the data within this time period, determined by dividing the difference between the mean values ​​of the preceding and following segments by the time interval; and the frequency of counts of consciousness disorder-related events within the window, such as the number of times the self-reported consciousness disorder index exceeds 0.8. Each type of calculation generates a new set of statistical features. Ultimately, the system will form multiple derived indicators based on the original matrix, used to construct a predictive model for consciousness disorder events.

[0029] Step 6: Data Storage and Output Interface Control: After the above processing, all structured data is stored in an individualized database table using the patient's unique ID as the primary key. The system provides a standardized output interface to push the latest daily updated dataset to subsequent modules. The interface supports data validation, historical data callback, and error log feedback. The default update frequency is once every 24 hours, which can be set to a higher frequency as needed. For the structured dataset of consciousness disorder features, pre-established symptom matching rules are used in conjunction with patient state classification methods to classify patient states into acute and remission phases, thus determining the patient's consciousness disorder classification result. This function, based on the structured dataset of consciousness disorder features, uses symptom matching rules and patient state classification methods to determine the consciousness disorder state at the current time point and its adjacent time periods, and outputs the consciousness disorder classification result accordingly. The complete process consists of four main steps: state determination, state trend smoothing, rule matching, and classification output.

[0030] Preliminary assessment of altered consciousness status: The system analyzes sampled data every minute, constructing a judgment matrix based on three characteristic indicators to determine whether it is in the "acute phase." The three characteristic indicators are as follows:

[0031] Severity of Consciousness Assessment: The assessment scale uses a standardized "Consciousness Impairment Rating Scale," which is entered by the patient's caregiver twice daily, once in the morning and once in the evening, via a mobile device. Each time, the caregiver slides their selection from "No Consciousness" (corresponding to a score of 0.0) to "Comatose" (corresponding to a score of 1.0) based on the current symptoms of consciousness impairment. The system maps this selection to a decimal value between 0 and 1, with precision retained to one decimal place. To achieve higher temporal resolution, the system uses linear interpolation to fill in the gaps between scores at the minute level. For example, if the caregiver scores 0.3 at 08:00 and 0.9 at 20:00, the system will fill in 720 sampling points every minute within this 12-hour interval with a linear increment of 0.000833, thus constructing a complete consciousness impairment severity curve. The value of each interpolation point is calculated from the time difference and score difference between two adjacent scores, ensuring a continuously usable intensity estimate at any given time point. Furthermore, the system has a preset intervention trigger threshold of 0.7. If the intensity score at a certain time point is greater than or equal to 0.7, then that time point is recorded as 1 point; otherwise, it is recorded as 0 points. This score will be used to determine whether to trigger intervention and will also serve as input in the calculation of the threshold determination and intensity decision model. The scoring mechanism is designed with convenience, individual subjective accuracy, and compatibility with subsequent algorithms in mind, ensuring that the degree of consciousness impairment is quantifiable, traceable, and responsive.

[0032] Neurological signal fluctuation score: The 30-minute moving amplitude of the average power of the high-frequency band of the EEG is used as the basic indicator. The system calculates the relative rate of change of power over the past 30 minutes every minute (the current value minus the average of the previous 30 minutes, then divided by the average of the previous 30 minutes). If the relative rate of change is greater than 0.2, it is scored as 1 point; otherwise, it is scored as 0 points. EEG power data is obtained through a head-mounted EEG acquisition device worn by the patient, with a sampling frequency of 64 times per second. The system performs bandpass filtering to extract the high-frequency band and averages it to the nearest minute.

[0033] The scores from the two items above are summed, with a maximum of 2 points. If the score at a certain point in time is 2 points, that point in time is marked as "acute phase"; otherwise, it is marked as "remission phase". This process runs in the system background once per minute, automatically outputting the current status flag.

[0034] Time trend correction mechanism:

[0035] To avoid misjudgments due to short-term fluctuations, the system introduces a sliding window trend smoothing method. The window length is 360 minutes, and it slides every 60 minutes to count the number of acute phase time points within the current window. A threshold ratio is set at 50%, meaning that if the number of acute phase time points within a window is greater than or equal to 180, the entire window is marked as an "acute phase"; otherwise, it is marked as a "remission phase". Each time point can belong to multiple sliding windows. The system determines the final state by aggregating the states of all windows covering that time point and using a majority voting principle. If two or more of the three sliding windows are identified as "acute phase", then the final state of that time point is "acute phase".

[0036] Symptom matching rules and calculation of consciousness impairment:

[0037] The system pre-defines five categories of consciousness impairment, each corresponding to a rule template. Each rule template consists of a feature list, with each feature corresponding to a condition and scoring criterion. Example rule (organic consciousness impairment) includes the following features: circadian rhythm disruption (more than 2 nighttime awakenings in the sleep log, or total daytime sleep exceeding 3 hours); disorientation (time or location orientation error rate greater than 0.5 in the assessment record); fluctuating attention (average reaction time fluctuation rate greater than 0.4 in interactive tasks); disorganized speech (proportion of abnormal semantic expressions greater than 0.3 in the speech record); and accompanying neurological signs (any one of the following during monitoring: hemiparesis, abnormal muscle tone, or nystagmus). One point is awarded for each condition met, with a maximum total score of 5 points for a single rule. The system performs a matching calculation with all rules daily based on feature data from the past 7 days, accumulating the score for each rule. The rule with the highest cumulative score is selected as the current primary category. If scores are the same, the rule with the higher score over the past 3 days is selected as the primary category, and the other is added as a secondary category.

[0038] The system outputs a classification structure and interface format: Consciousness state assessment and consciousness disorder identification results every 24 hours, including: minute-by-minute consciousness state markers (e.g., awake, drowsy, delirious, minimally conscious, vegetative state, coma, etc.); consciousness state trend labels every 360 minutes (e.g., increased fluctuations, improved clarity of consciousness, persistent low responsiveness, etc.); the currently identified primary and secondary consciousness disorders; detailed scores for each rule template (listing the matching status and score for each feature); and a summary of key feature values, including daily maximum, minimum, and average values ​​for parameters such as orientation disorder score, attention maintenance index, language confusion index, level of spontaneous activity (e.g., steps per unit time or number of interactions), and EEG fluctuation rate. All the above outputs are encapsulated in a structured data format (e.g., JSON or CSV) for subsequent intervention module reading and adjustment of intervention strategy intensity. Simultaneously, the system supports historical record retrieval and dynamic display of rule parameters, ensuring a closed-loop feedback between state identification and intervention optimization. All threshold parameters (such as orientation error frequency 0.5, attention fluctuation rate 0.4, EEG fluctuation 20%, sliding window 360 minutes, etc.) are based on the statistical distribution of characteristic data from 1,000 previous clinical patients with consciousness disorders and have been set after being reviewed and approved by a team of neurological clinical experts, thus possessing scientific validity and universality. At the same time, the system reserves a manual fine-tuning interface to support parameter adjustment and personalized adaptation according to individual differences and scenarios.

[0039] This approach, by combining structured data on characteristics of consciousness disorders with symptom matching rules, achieves accurate classification of patients' current state of consciousness and precise identification of consciousness disorders, significantly improving the targeting and individual adaptability of intervention decisions. By introducing a multi-dimensional indicator comprehensive judgment mechanism and a time trend smoothing algorithm, the system effectively reduces the risk of misjudgment caused by short-term fluctuations (such as momentary slow reaction time or loss of attention), improving the stability and continuity of consciousness state identification. Furthermore, the matching rules are constructed based on a large sample of clinical patients with consciousness disorders, ensuring high clinical reliability of the classification results and providing solid data support and an algorithmic foundation for the personalized adjustment and dynamic optimization of subsequent intervention plans.

[0040] Individualized features of frequency and duration of consciousness disorder symptoms were extracted from the classification results of consciousness disorder. Intervention priority ranking logic was used to determine the patient's core symptom intervention needs, resulting in a personalized symptom priority vector.

[0041] This module aims to quantitatively analyze the frequency and duration of abnormal consciousness states in patients based on the classification results of consciousness disorders. It further integrates this with symptom intervention priority ranking logic to establish a structured intervention target ranking mechanism for individuals. The entire process includes three stages: consciousness disorder feature extraction (such as disorientation, attentional fluctuations, and language disorder), symptom pattern recognition (identifying consciousness state patterns such as periodic fluctuations, acute exacerbations, or persistent low responsiveness), and priority vector generation. Priority vectors are generated, and multiple factors are weighted and ranked according to symptom severity, scope of impact, and intervention feasibility to ultimately form an individualized intervention strategy reference sequence. Stage 1: Individualized Consciousness Disorder Feature Extraction. The calculation method for the frequency of consciousness disorders: The system acquires minute-level consciousness state data from the patient over the past 7 consecutive days, labeling each minute as either "abnormal state" (such as drowsiness, delirium, lethargy, minimal consciousness, vegetative state, etc.) or "awake state." By traversing each minute, the system identifies the point in time when the state switches from "awake state" to "abnormal state," and each switch is recorded as a consciousness disorder event. For example, if the status label shows a switch from "awake to abnormal" 5 times out of 10080 time points, the frequency of altered consciousness episodes is 5 times per 7 days. The frequency of altered consciousness episodes is categorized into three levels based on the number of episodes per 7 days: High frequency: ≥5 times per 7 days; Medium frequency: 3 to 4 times per 7 days; Low frequency: less than 3 times per 7 days. The system records the start and end times of each abnormal state of consciousness (i.e., abnormal period) and calculates the duration of each abnormal period in minutes. For example, if an abnormal state starts at minute 3000 and ends at minute 3300, its duration is 300 minutes. The system also calculates the average duration of all abnormal periods over the past 7 days. The duration of altered consciousness episodes is categorized as follows: Long duration: ≥240 minutes; Medium duration: 120 to 239 minutes; Short duration: less than 120 minutes. After the system completes the classification of the two indicators mentioned above, it will mark the current patient as a "frequency-duration" composite type, such as "high frequency long duration" or "medium frequency short duration", for the precise formulation of subsequent individualized intervention strategies.

[0042] Intervention Priority Ranking Logic. Intervention Goal Definition: This system sets three core intervention goals for the management of consciousness disorders: reducing the frequency of consciousness disorder attacks, shortening the duration of a single attack, and reducing the degree of consciousness disorder during an attack (such as drowsiness, minimal consciousness, vegetative state, coma, etc.). Based on the "frequency-duration" composite type identified in the first stage, the system establishes a matching relationship between 9 symptom types and their corresponding intervention priority vectors. The priority is 1 as the highest and 3 as the lowest. The specific templates are as follows: High frequency long duration: [1, 2, 3]; High frequency medium duration: [1, 3, 2]; High frequency short duration: [1, 3, 2]; Medium frequency long duration: [2, 1, 3]; Medium frequency medium duration: [2, 3, 1]; Medium frequency short duration: [3, 2, 1]; Low frequency long duration: [2, 1, 3]; Low frequency medium duration: [3, 2, 1]; Low frequency short duration: [3, 2, 1]. After the system classifies the frequency and duration of a patient's altered consciousness episodes, it automatically uses the corresponding priority template as the initial intervention strategy ranking for that individual. For example, if a patient is classified as "high-frequency, long-duration," the system will select a priority vector [1, 2, 3], meaning the priority intervention goal is to reduce the frequency of episodes, followed by shortening the duration of each episode, and finally reducing the severity of impairment during an episode. This priority mechanism provides a basis for the intelligent scheduling and personalized resource allocation of subsequent intervention measures.

[0043] Priority vector fine-tuning mechanism

[0044] If the system detects that a patient has severe comorbid symptoms within the past 3 days, the system will fine-tune the position of the original intervention priority vector. Specifically, the adjustment rule is as follows: intervention goals with a score of 2 in the original vector are adjusted to 1, and those originally 1 are adjusted to 2, ensuring that the consciousness impairment characteristics associated with the current significant comorbid symptoms are given priority in response and intervention. In terms of output structure, the system outputs the final processing result as a three-dimensional vector, with each dimension representing the ranking of the three types of intervention goals: reducing the frequency of attacks, shortening the duration of attacks, and alleviating the severity of attacks. For example, [2, 1, 3] indicates that the highest priority intervention goal is to shorten the duration of each consciousness impairment attack, followed by reducing the frequency of attacks, and finally alleviating the severity of impairment during attacks. This priority vector is generated daily and updated every 24 hours, and is used for weighting the intensity curve and triggering decision control nodes in subsequent intervention plans. All threshold settings in intervention decisions are based on clear quantitative evidence and three authoritative data sources to ensure the scientific validity and repeatability of the standards. First, the system collected continuous state monitoring data from 1000 patients with disorders of consciousness, calculating the median and upper and lower quartiles of key indicators (such as frequency of episodes, duration of a single abnormal state, orientation score, and attention fluctuations) to clarify the characteristic distribution range in actual clinical practice. Second, the system referenced the classification thresholds recommended by international clinical assessment standards for disorders of consciousness (such as RASS and CAM-ICU) for indicators such as frequency of episodes (e.g., ≥2 times per day for severe fluctuations) and duration (e.g., ≥240 minutes for long-term disorders), and compared and corrected the system's internal parameters accordingly. Finally, through the annotation, verification, and discussion of real clinical data by neurology and geriatric psychiatry experts, a multi-round parameter optimization mechanism was constructed to determine the cutoff values ​​for each graded variable. For example, defining a frequency of disorders of consciousness greater than or equal to 5 times every 7 days as "high frequency" and a single duration of greater than or equal to 240 minutes as "long-term" is based on the overlap range between the 75th percentile of the sample statistics and the recommended values ​​of international standards, possessing sufficient statistical support and medical consistency.

[0045] By extracting individualized features of the frequency and duration of consciousness disturbances from the classification results, and combining this with intervention priority ranking logic to generate personalized symptom priority vectors, this approach can accurately identify the main intervention needs of patients at different stages, realizing the transformation of intervention strategies from a uniform template to individual customization. Based on a multidimensional consciousness state data-driven discrimination mechanism, this method significantly improves the targeting and response speed of intervention plans, avoids inappropriate resource allocation and poor intervention effects, enhances the system's intelligent adaptability and individual difference recognition capabilities, and provides an efficient and adjustable decision support tool for the continuous optimization of consciousness disturbance management.

[0046] Based on personalized intervention priority vectors for consciousness disorders and patients' historical data on abnormal states of consciousness, the system constructs a preliminary intervention time point distribution logic adapted to the patient's specific attack patterns, generating an initial daily intervention schedule. This process includes five main steps: high-incidence probability modeling of consciousness disorders, time period weight calculation, priority ratio conversion, intervention time point distribution configuration, and schedule output and update. The first step, high-incidence probability modeling of consciousness disorders: The system extracts 1440 time points of consciousness (abnormal or lucid periods) from the patient's minute-level consciousness state data over the past 14 days, dividing the daily timeline into 24-hour segments of 60 minutes each. The system counts the cumulative number of minutes of "abnormal states" within each hourly segment over 14 days. For example, if a patient experiences 420 abnormal states between 9:00 AM and 10:00 AM, the probability of abnormality during that period is 420 divided by 840, resulting in 0.5 (840 being the maximum possible number of minutes within that period over 14 days). All hourly segments undergo the same processing, ultimately forming the relative probability of consciousness disorders occurring within a 24-hour period. The second step is time period weight calculation: To achieve normalization, the system standardizes the time period with the highest probability of occurrence as weight 1, and scales the remaining time periods proportionally to generate a 24-dimensional time period weight vector, which serves as a reference for subsequent intervention point configuration. The third step is priority ratio conversion logic: The system reads the consciousness disorder intervention priority vector generated by the preceding module. For example, [1, 2, 3] represent reducing the frequency of attacks, shortening the duration, and alleviating the severity of the disorder, respectively, corresponding to intervention resource ratios of 60%, 30%, and 10%. This ratio was derived by an expert team based on intervention response data from 100 typical patients, balancing intervention effectiveness and resource allocation. Based on this, the system determines a resource allocation structure of 48 fixed intervention time points per day (one every 30 minutes, covering the entire day). The fourth step is intervention time point distribution configuration: Taking priority 1 as an example, 60% of the time points, i.e., 29 points, need to be configured. The system prioritizes selecting the hour segment with the highest probability of attacks, sorting and accumulating them by weight until the target distribution weight of 60% is met (e.g., the target weight is 5.76 units), and evenly distributing the 29 intervention points within the corresponding time period. The allocation logic for priority 2 and priority 3 is similar, configuring 14 and 5 time points respectively. If a certain hourly segment is selected by multiple priority targets simultaneously, the system automatically shortens the intervention interval for that segment to 20 minutes to ensure intervention density and coverage intensity of priority targets. The fifth step is the timetable output mechanism: the system sorts the 48 intervention time points by minute, with each point accompanied by three pieces of information: specific time (e.g., 540 minutes, i.e., 9:00 AM), intervention target identifier (e.g., frequency control), and corresponding time period weight value (e.g., 0.85), outputting it in a structured data format for subsequent reading and use by the "intervention intensity allocation module." This timetable is updated daily; the system automatically runs the algorithm 10 minutes before midnight each day, generating an updated table based on the latest 14 days of data and archiving logs to support backtracking and manual verification.All parameter settings are derived from clinical research and expert consensus: 48 time points per day represent the optimal cognitive load limit for patients; 30-minute intervals correspond to the rhythmic changes in consciousness; priority ratios are based on empirically optimal configurations; and a 14-day window reflects the stability of behavioral patterns. This timetable generation algorithm, through distributed weight evaluation and goal-driven configuration, constructs a highly personalized and rhythmically matched intervention time distribution for consciousness disorders, possessing both engineering feasibility and clinical compatibility.

[0047] For the initial intervention timetable for consciousness disorders, a time interval subdivision logic was adopted to divide the timetable into smaller time units. At the same time, the peak period of consciousness disorders was identified as the core intervention point through key node extraction technology, resulting in a detailed intervention time framework.

[0048] After the initial intervention schedule for disorders of consciousness is generated, to further improve the accuracy of the intervention rhythm and the efficiency of time period matching, the system introduces time interval subdivision logic and key node extraction technology to refine and reconstruct the daily intervention schedule, constructing a refined intervention time framework that conforms to the individual's rhythm of disorders of consciousness. This technical process consists of five steps, covering time unit division, abnormal state density matrix construction, high-risk threshold calculation, key node extraction, and structured intervention point allocation. The first step is time unit subdivision: the original intervention schedule is based on 48 points, each lasting 30 minutes. To improve the sensitivity of the time-series response, the system divides the 1440 minutes of the day into 144 time units, each lasting 10 minutes, corresponding to numbers 1 to 144, covering 00:00 to 00:10, 00:10 to 00:20, ... up to 23:50 to 00:00. The 10-minute granularity is derived from system performance testing and the clinical response window for disorders of consciousness intervention, meeting the requirements of refined monitoring while avoiding the burden on patients caused by frequent interventions. The second step is the construction of the density scoring matrix: The system analyzes the patient's minute-level consciousness status data from the past 7 days, counting the frequency of "abnormal states of consciousness" (such as drowsiness, delirium, minimal consciousness, vegetative state, coma, etc.) within each 10-minute segment, assigning 1 point for each occurrence. The final score range for each 10-minute segment is 0 to 98 (7 days multiplied by the maximum probability of 14 occurrences). For example, if the time segment numbered 78 (13:00-13:10) accumulates 14 abnormal states, its density score is 14. Based on this, the system constructs a 144-dimensional density vector as the probabilistic basis for identifying high-risk time periods. The third step is the calculation and classification of peak thresholds: The system sorts the density vectors and selects the top 20% of time periods as candidate high-risk time periods, i.e., the first 29 time periods. The lowest score among these 29 segments is then calculated as the lower threshold. For example, if the score of segment 29 is 12, all time periods with scores greater than or equal to 12 are marked as "high-risk segments." This 20% figure is derived from multiple rounds of clinical empirical studies, showing that over 80% of consciousness disorders occur within approximately 15%-25% of the time period, serving as the basis for determining key intervention windows. Through the above steps, the system can effectively identify and focus on critical time points with high incidence of consciousness disorders, supporting more intensive and precise intervention rhythm settings, and providing a refined input basis for subsequent intervention target overlay, intensity adjustment, and temporal weighting.

[0049] Next, the system performs continuous merging on time periods marked as high-risk for altered consciousness. If the interval between adjacent high-risk periods does not exceed one 10-minute interval (i.e., within 10 minutes), they are merged into a continuous "core node segment." After merging, the system checks the total duration of each core node segment. If it is less than 20 minutes (i.e., less than two consecutive high-risk periods), it is considered to have insufficient clinical intervention value and is removed from the configuration. Ultimately, the system retains a maximum of 5 core node segments for concentrated deployment of high-frequency intervention time points. The fourth step is the time point structure configuration algorithm: the system expands the number of daily intervention time points from the initial 48 to a maximum of 72 points, prioritizing allocation to core node segments within resource limits. Each core segment is configured with 2 to 4 intervention points based on its duration, with a default interval of 10 to 15 minutes. For example, if a core segment is located between 08:40 and 09:20, covering four 10-minute intervals, the system may set three time points: 08:40, 09:00, and 09:20 to ensure the time density of intervention responses during high-incidence periods. In non-core segments, the system focuses intervention on the "medium-risk segment" where the density score is between 30% and 60%. These periods represent potential peak times for secondary attacks, and the system allocates auxiliary time points on an hourly average, with a maximum of one per hour, to moderately intervene and control the spread of abnormal conditions. For the remaining time periods with a density score below 60%, i.e., the low-risk area, the system does not allocate any time points to avoid unnecessary interventions and wasted system resources, and also to reduce the daily burden on patients. The fifth step is the generation and formatted output of the intervention timeframe: The system ultimately generates a structured list of no more than 72 intervention time points. Each point contains the following key information fields: time number (1 to 144, corresponding to one of the 144 10-minute segments throughout the day), point type ("core point" or "auxiliary point"), high-risk segment identifier (if belonging to a core segment, the segment number is noted), density score (reflecting the historical probability of abnormal consciousness at that time point), and ranking weight (serving as dynamic weight input for the subsequent intervention intensity adjustment module). This timeframe is automatically refreshed daily before 00:00 and simultaneously output to the system's intervention intensity scheduling module and patient terminal devices, achieving 24 / 7 closed-loop feedback control. This refined intervention timeframe, by accurately identifying and strengthening coverage of high-incidence periods of consciousness disorders, combined with a density-driven point distribution strategy, achieves an effective combination of high temporal resolution, high intervention targeting, and low system burden, ensuring the scientific, adaptable, and operable nature of the intervention rhythm in individualized consciousness disorder management.

[0050] This technical solution effectively improves the precision and targeting of intervention rhythm by subdividing the initial intervention schedule for consciousness disorders into time intervals and combining key node extraction technology to identify peak periods of consciousness disorders as core intervention points. By dividing the daily timeline into 144 10-minute units and accurately identifying high-incidence time periods based on the symptom density distribution over the past 7 days, the system can dynamically lock onto the core periods with the highest probability of consciousness disorders, centrally allocate intervention resources, and achieve a high degree of matching between intervention timing and the rhythm of attacks. Simultaneously, this method, while ensuring response sensitivity, rationally controls the total number of intervention points, avoiding resource waste and over-intervention, providing a high-resolution time control mechanism for personalized and precise intervention, and enhancing the system's adaptability and clinical applicability.

[0051] From the detailed intervention timeframe, combined with the intensity decision model, the initial stimulus intensity is calculated, and the stimulus intensity at the intermediate point between adjacent key nodes is calculated using the intensity value interpolation formula, thus obtaining the preliminary intervention intensity curve.

[0052] After completing the detailed intervention timeframe construction, to ensure the continuity and precision of stimulus intervention, the system introduces an intensity decision model to calculate the initial stimulus intensity value for each core intervention node. Interpolation methods are then used to fill in the intensity values ​​at intermediate points between nodes, thus forming a continuous preliminary intervention intensity curve over the entire 1440 minutes of the day. This process can be broken down into four consecutively executed technical steps: core node intensity calculation, normalized weight allocation, interpolation segment division, and interpolation process execution.

[0053] Step 1: Initial Stimulus Intensity Calculation for Core Nodes. The system first identifies all marked core intervention nodes. Each node has three input features: Density Score: The cumulative number of "acute phase" episodes occurring in the past 7 days within the 10-minute timeframe of this node, ranging from 0 to 98. The system divides this value by 98 to obtain the normalized density score D, ranging from 0 to 1. Symptom Intervention Priority: The primary intervention target for this node, with a priority of 1, 2, or 3. The system assigns weight values ​​W to the priorities, which are 0.6, 0.3, and 0.1, respectively. These weights are determined by an expert group through weight sensitivity analysis to ensure focused intervention. Time-Segment Risk Weight: The probability weight of consciousness impairment episodes within the hourly timeframe of the node, calculated previously, ranging from 0 to 1, denoted as R. The system uses these three values ​​to calculate the initial stimulus intensity S, employing a weighted average: S = D multiplied by 0.5, plus W multiplied by 0.3, plus R multiplied by 0.2. In the above parameter coefficients, 0.5, 0.3, and 0.2 represent the weight distribution of historical density, intervention target priority, and time risk, respectively. This ratio was determined iteratively by the system in response to 500 patients, balancing time dependence and individual response differences. The resulting S-value is limited to the range of 0.2 to 1. If the calculated result is less than 0.2, the system automatically adjusts it to 0.2; if it is greater than 1, the system sets it to 1.

[0054] Step 2: Interpolation Segment Definition. The system arranges all core intervention nodes chronologically, marking their positions within the 144 time segments of the day. Then, it sequentially extracts all empty segments between adjacent nodes, defining each segment as an interpolation interval. For example, if the first node is in segment 48 (8:00 AM) and the second node is in segment 60 (10:00 AM), the interpolation interval includes segments 49 to 59, a total of 11 points. The system sets the minimum length of the interpolation interval to 2 segments (20 minutes) and the maximum length to 9 segments (90 minutes); intervals exceeding this are interpolated in segments. This interval was determined by research on clinical intervention response lag and neural recovery mechanisms, balancing continuity and physiological tolerance.

[0055] Step 3: Interpolation Calculation Process. Within each interpolation segment, the system uses a linear interpolation method to calculate the stimulus intensity value for each intermediate segment. Let the intensity of the starting node be A, the intensity of the ending node be B, and the number of intermediate interpolation points be N. Then, the intensity value corresponding to the nth interpolation point (counting from 1) is: the starting intensity plus the difference multiplied by the proportion of the nth point to the segment length, that is: A plus (B minus A), then multiplied by n and divided by (N plus 1). For example, if A is 0.6, B is 0.9, and N is 3, then the intensities of the three interpolation points are: Point 1: 0.675; Point 2: 0.75; Point 3: 0.825. This algorithm ensures that the interpolated values ​​change continuously and smoothly between the preceding and following nodes, avoiding the system shock caused by abrupt interventions.

[0056] Step 4: Intensity Curve Integration and Output. The system aggregates the intensity values ​​of all core nodes and the intermediate point intensity values ​​generated in the interpolation segments to form a 1440-minute intervention intensity sequence for the entire day. Intensity values ​​for unspecified time periods in the sequence are set to 0 by default, indicating a non-intervention state. Finally, the system generates a set of daily updated preliminary intervention intensity curve datasets. Each data point includes its time location, intervention intensity value, and whether it is a core point marker. The system allows users or doctors to set maximum intervention intensity limits, minimum response stimulus thresholds, and restrictions on intervention during specific time periods to control the overall shape of the curve and enhance individualized adjustment capabilities.

[0057] This technical solution combines a refined intervention timeframe with an intensity decision model to accurately calculate the initial stimulus intensity based on node characteristics. A linear interpolation method is then used to fill in the intermediate intensity values ​​between adjacent key nodes, thereby constructing a complete, continuous, and rhythmic preliminary intervention intensity curve. This method significantly improves the dynamic matching capability of the intervention plan, ensuring sufficient stimulus response intensity at key time points while achieving a natural transition in intensity in non-core areas, avoiding physiological shocks or systemic interference caused by abrupt interventions. Simultaneously, the intensity curve is controllable and adjustable, providing a stable foundation for subsequent smoothing and real-time feedback correction, enhancing the accuracy, continuity, and effectiveness of personalized intervention plans.

[0058] For the initial intervention intensity curve, boundary condition constraint technology is used to set upper and lower limits for intervention intensity and time. At the same time, the importance parameters of time points are adjusted by the smoothness weight allocation method to obtain a smooth intervention implementation plan.

[0059] To ensure the feasibility and rhythmic stability of the initial intervention intensity curve, after generating the basic intensity value sequence, the system employs boundary condition constraint techniques to set upper and lower limits for the stimulus intensity and time distribution, and introduces a smoothness weighting method to control the continuity and rhythm of curve changes. The entire process consists of five steps: intensity value boundary constraints, time interval boundary limits, variation amplitude constraint processing, importance weight calculation, and local smoothing correction.

[0060] Step 1: Setting the upper and lower limits of intervention intensity. The system iterates through each time point in the initial intervention curve, performing the following intensity boundary checks: The minimum intensity limit is set to 0.2. If the intensity at a point is less than 0.2, its value is set to 0, indicating that no stimulation is needed at that time point; this value is set with reference to the minimum threshold for stimulation effectiveness, derived from clinical research showing that neural stimulation below this value does not have a measurable physiological effect. The maximum intensity limit is set to 1. If the intensity at a point exceeds 1, it is corrected to 1 to prevent energy from exceeding the maximum tolerance range of the equipment. The minimum effective difference is set to 0.05. If the intensity difference between two adjacent points is less than 0.05, the smaller value will converge towards the larger value to make its change significant, avoiding noise level disturbances affecting system behavior.

[0061] Step 2: Limiting the intervention time interval. The system iterates through all time series with non-zero intensity points, calculates the time difference between adjacent intervention points, and executes the following strategy: If the interval between two points is less than 10 minutes, it indicates that the points are too densely packed. The system retains the point with higher intensity and deletes the other point to ensure that the physiological system has sufficient recovery space within the response cycle. If the interval between two points is greater than 60 minutes, it indicates that there is a possible rhythm gap. The system automatically inserts a transition point at the midpoint of the interval, with the time position being the middle 10-minute segment and the intensity value being the average of the two points before and after, to ensure the continuity of stimulus coverage. The above 10-minute and 60-minute parameters are derived from the statistical results of the neural regulation recovery cycle (lowest stimulus re-entry interval) and the smallest unit of symptom rhythm (regulation period within a single hour), respectively.

[0062] Step 3: Intensity jump amplitude limitation processing. To prevent sudden increases or decreases in the curve, the system performs intensity difference detection on each pair of adjacent points. If the intensity jump amplitude is greater than 0.3, the system activates a step-by-step transition mechanism, automatically inserting 1 to 2 transition points in the middle, with the time interval evenly distributed, and the total difference amortized to the middle points. For example, if the intensity at the starting point of a segment is 0.3 and the intensity at the ending point is 0.9, with a difference of 0.6, then two points are inserted in the middle, with intensities of 0.5 and 0.7 respectively, achieving a linear and smooth transition. 0.3 is the maximum allowable jump amplitude, an acceptable jump threshold derived from the analysis of patient heart rate and EEG response data.

[0063] Step 4: Time Point Importance Weighting. The system calculates a smoothing control weight for each intervention point, with a value range of 0 to 1. It consists of three parts: Rhythm Change Rate Weight (P1): The sum of the intensity differences between the current point and its preceding and following points divided by 2. If the result is greater than 0.2, P1 is 1; otherwise, it is scaled proportionally. Symptom History Overlap Weight (P2): The frequency of acute phases occurring in the 10-minute segment containing the current point in the past 7 days divided by 98. A larger value indicates a higher overlap with high-incidence periods, and a higher weight. Node Type Weight (P3): Core nodes have a weight of 1, interpolation points have a weight of 0.5, and newly added transition points have a weight of 0.3. The overall weight W is P1 multiplied by 0.5, plus P2 multiplied by 0.3, plus P3 multiplied by 0.2, with the result rounded to two decimal places.

[0064] Step 5: Smoothing the Implementation and Intervention Plan Output. After calculating the importance weights of all points, the system performs a three-point moving average smoothing process on points with a W value less than 0.5. The processing logic is as follows: the current value is replaced by a weighted average of the intensity values ​​of the current point and the two points before and after it, with weights of 0.25, 0.5, and 0.25 respectively. Points with a weight greater than or equal to 0.5 retain their original values, preserving their responsiveness in the curve. The final output intervention implementation plan is a structured array of length 144, corresponding to 1440 minutes with one time point every 10 minutes. Fields include time number, intensity value, whether it is a core point, and smoothing weight value. The system reconstructs the intervention implementation plan every 24 hours to adapt to symptom evolution and updated feedback data.

[0065] This technical solution introduces boundary condition constraints based on the initial intervention intensity curve, setting upper and lower limits for intensity and time intervals for each intervention point. Combined with a smoothness weighting method, it dynamically calculates the importance of each time point based on rhythm variability, symptom overlap, and node type. This enables partitioned optimization and local adjustment of the curve, effectively eliminating abrupt changes, overly dense interventions, and gaps. This method not only improves the continuity and physiological rhythm matching of the intervention curve but also ensures the complete preservation of intervention intensity at key nodes, significantly enhancing the controllability, rhythm stability, and individual adaptability of the execution plan. It provides a high-quality temporal foundation for subsequent feedback regulation and closed-loop control.

[0066] Based on the smooth intervention implementation plan, feedback data on the relief of patients' consciousness impairment symptoms during the intervention period are collected in real time. If the feedback data deviates from the preset threshold range, the rule base parameters are adjusted through dynamic rule update technology. At the same time, combined with interpolation error correction and curve continuity verification, the intervention schedule and stimulus intensity curve are updated to determine the optimized intervention implementation plan.

[0067] This technical solution is based on a real-time feedback control mechanism. Through five processes—feedback acquisition, anomaly identification, dynamic parameter correction, local interpolation adjustment, and curve continuity verification—it achieves adaptive optimization and upgrading of the initial execution plan. The specific implementation process is as follows:

[0068] I. Feedback Data Collection and Standardized Calculation: The system automatically collects patient and caregiver pain scores (range 0 to 10) every 10 minutes via wearable devices, along with auxiliary indicators such as blood pressure changes, skin conductivity, and sleep interruption markers. Subjective score data is used to determine the level of improvement in consciousness impairment, and is standardized according to the following criteria: scores of 0 to 2 are classified as no relief, with a standard value of 0; scores of 3 to 5 are classified as mild relief, with a standard value of 0.3; scores of 6 to 8 are classified as moderate relief, with a standard value of 0.6; and scores of 9 to 10 are classified as significant relief, with a standard value of 1. Standardized feedback values ​​are generated at each time point and incorporated into the feedback data sequence.

[0069] II. Feedback Deviation Judgment Logic and Calculation Rules: The system sets a target relief value T for each patient, categorized by patient type: mild consciousness impairment (T = 0.6); moderate consciousness impairment (T = 0.7); severe consciousness impairment (T = 0.8). The system compares the feedback value F at the current time point with the target value T. A deviation is determined if either of the following two conditions exists: F values ​​are less than T minus 0.2 for three consecutive time points; or F is less than 0.4 at any time point, accompanied by abnormal auxiliary indicators collected by the device.

[0070] III. Dynamic Rule Base Parameter Update Process: After a deviation judgment is triggered, the system updates the following parameters of the intervention rule base: the rule number records the current intervention time period; the symptom priority value is increased by 1 level, for example, from 0.3 to 0.6; the historical symptom density value is increased by 20%; the current node intensity correction factor is set to 1.2, that is, the intensity of the same situation in subsequent interventions is multiplied by 1.2; the deviation sample is added to the "abnormal pattern template" sub-library, and the template validity period is set to 7 days. After the expiration, the weight is reduced.

[0071] IV. Interpolation Error Correction Mechanism and Calculation Logic: If the deviation area spans multiple intervention nodes, the system performs interpolation correction on the intermediate segment between adjacent intervention points A and B: Intensities A and B are Sa and Sb, respectively; the interpolation segment contains N intermediate points; the intensity at the nth position of each intermediate point is Sa plus Sb minus Sa, multiplied by n, divided by N plus 1; if the feedback deviation level is severe (i.e., the feedback value is less than 0.3), all interpolation results are multiplied by an additional factor of 1.2 to 1.4 based on the original formula, set according to the deviation level. The system ensures that a single correction does not exceed 30% of the original value to guarantee the physiological tolerance of the intervention stimulus.

[0072] V. Curve Continuity Verification Calculation Method: After correction, the intervention intensity curve undergoes integrity verification, primarily checking the following: Intensity Abrupt Value Detection: Any intensity difference exceeding 0.3 between two adjacent points is defined as an abrupt change. The system lowers the larger value below the critical value and inserts two equally spaced increasing (or decreasing) points for smoothing. Intervention Gap Identification: Any consecutive blank (intensity 0) segment exceeding 6 points (i.e., 60 minutes) is defined as a gap. The system inserts a point with an intensity value of 80% of the previous non-zero segment in the middle position to ensure no loss of intervention rhythm. Local Extreme Value Verification: If the intensity value of any non-core node is greater than the average value on both sides plus 0.4, the system identifies it as a false peak and adjusts it to match the average values ​​of the left and right points.

[0073] VI. Final Optimization and Implementation Plan Generation: After completing the above processing, the system reconstructs the intervention implementation plan, containing 144 time points. Each time point includes the following fields: time number (1 to 144); final stimulus intensity (range 0.2 to 1); whether it is a dynamically corrected node; current node weight value (for subsequent optimization); response template flag (whether it is included in the negative feedback template). This implementation plan will be reconstructed every 24 hours and will be updated immediately when the patient reports significant abnormalities, ensuring that the stimulus and symptom evolution remain synchronized.

[0074] The preset threshold range was determined based on a large number of clinical samples through statistical analysis and expert consensus, and is used to judge the effectiveness of intervention feedback. The determination process includes three steps: First, the system collects data on the improvement of consciousness disorders in patients with different levels of consciousness before and after intervention, and calculates the median and the 25th and 75th percentiles for each score sequence, serving as the baseline remission level and upper and lower limits, respectively. Second, it sets expected remission values ​​for different symptom levels, for example, 0.6 for mild, 0.7 for moderate, and 0.8 for severe. Finally, the expert group performs cluster analysis on different behavioral response data to assess the actual remission probability under the target stimulus intensity, determining the acceptable deviation range for each symptom type as the target value minus 0.2 to the target value plus 0.2. Ultimately, the system incorporates this threshold range into the feedback judgment logic to judge in real time whether the feedback data deviates from the standard, ensuring that the judgment process is both scientifically based and adaptable to individual patient differences.

[0075] This technical solution achieves closed-loop regulation and adaptive optimization of the intervention process by introducing a real-time feedback mechanism and dynamic rule update technology. During the intervention, the system continuously collects subjective relief scores and physiological response data from both the patient and caregiver, comparing them with preset quantitative threshold ranges. If deviations are detected, dynamic adjustments to rule parameters, intervention curve interpolation correction, and continuity verification are triggered, thereby updating the intervention schedule and stimulus intensity curve. This method not only improves the system's sensitivity and response speed to individualized changes in condition but also effectively avoids the problem of insufficient adaptation of fixed intervention protocols in complex cases. By continuously optimizing the execution path, the stability and efficiency of the intervention effect are improved, realizing a data-driven precision treatment strategy centered on patient feedback, and enhancing the system's intelligence, security, and practical clinical application value.

[0076] For the optimized intervention implementation plan, a personalized threshold setting technique is adopted to adjust the triggering conditions for intervention of consciousness disorders according to individual differences. At the same time, the latest feedback data is integrated into the intervention intensity curve through real-time data fusion method to obtain the final dynamic intervention implementation logic.

[0077] To achieve precise and dynamic intervention and control of consciousness disorders, this technical solution starts with determining individual thresholds and combines them with real-time feedback data to construct the final intervention triggering logic and intensity adjustment mechanism. The overall process consists of three core calculation stages: threshold setting calculation, integrated feedback adjustment, and dynamic output logic construction.

[0078] I. Personalized Threshold Setting Calculation Process: The system first extracts the individual response characteristics of each patient from historical feedback data and calculates the personalized intervention trigger threshold. The specific steps are as follows: Sample Extraction: Select valid feedback data from 144 time points per day over the past 7 days, remove missing and outlier points, and form a standard scoring sequence with a total data volume of no less than 720 points. Median and Variation Range Calculation: Calculate the median, 25th percentile, and 75th percentile of the sequence. For example, if the median is 0.65, the 25th percentile is 0.55, and the 75th percentile is 0.75, the baseline remission range is 0.55 to 0.75. Matching Target Remission Level: Based on the type of consciousness impairment determined by the patient during registration, the system assigns a remission target, for example, 0.8 for severe cases, 0.7 for moderate cases, and 0.6 for mild cases. Set personalized trigger values: If the median value is less than or equal to the target value by 0.1, the trigger value is set to the target value; if the median value is more than or equal to the target value by 0.1, the trigger value is set to the median value minus 0.05; if the median value is more than or equal to the target value by 0.2, the trigger value is set to the median value plus 0.1. The resulting personalized trigger threshold V (generally between 0.5 and 0.85) is used to determine whether to initiate intervention.

[0079] II. Real-time data fusion and intervention intensity adjustment calculation: Every 10 minutes, the system collects the current feedback value F once and dynamically adjusts the intervention intensity S at the current time point according to the following rules: If F is higher than V + 0.1: the current stimulus intensity decreases by 10%, that is, S is set to the original value multiplied by 0.9; If F is lower than V - 0.2: the current intensity increases by 15%, that is, S is multiplied by 1.15; If F is between V - 0.1 and V: the current intensity remains unchanged, but the node is marked with a "sensitive label" for subsequent reinforcement processing; If F is lower than V - 0.2 for 3 consecutive time points: the current node is marked as a "high priority response point" and given a time importance weight of more than 0.8 in subsequent curve construction.

[0080] III. Dynamic Intervention Logic Output Construction Process: Combining the personalized trigger value V and the intensity S adjusted by real-time feedback, the system generates a final intervention instruction set for each 10-minute segment, including time numbers: 1 to 144, corresponding to 10-minute segments throughout the 1440 minutes of the day; output intensity value S: the corrected value, limited to between 0.2 and 1; trigger flag: whether the current time point F is lower than V, if so, the trigger flag is set to "yes", otherwise "no"; priority label: divided into three categories—high, medium, and low—based on the current feedback and historical deviation, with values ​​of 0.8, 0.5, and 0.3 respectively; fusion correction label: recording whether the point is a fusion update node, marked as "Y" or "N". Each record is stored in the execution plan table, and the system automatically completes a total plan update every 24 hours. In special cases, if a rapid deviation trend is detected, an immediate update logic can be triggered. The system uses this dynamic plan as the dominant intervention sequence in the next cycle to achieve adaptive closed-loop control of feedback-adjustment-re-intervention.

[0081] The intervention intensity S at the current point in time refers to the stimulus intensity value calculated by the system based on individual feedback status and predetermined intervention strategies within a specific time period. It is the core parameter for implementing dynamic intervention. This value comprehensively considers the personalized trigger threshold V, real-time feedback value F, and feedback trend to ensure that the intervention intensity meets clinical efficacy requirements without exceeding the individual's tolerable range. In practice, if F is significantly lower than V, the system automatically increases S to enhance the intervention response; if F is significantly higher than V, it indicates that the current intervention has produced a good effect, and S is adjusted accordingly to prevent overstimulation. The value of S is strictly limited to between 0.2 and 1, and is fine-tuned through multi-point feedback trends and node priority labels to ensure the stability and individual adaptability of intensity adjustments at different nodes. Ultimately, S not only reflects the intensity of intervention demand in the current state but also serves as a fundamental variable for constructing and updating the dynamic intervention curve in real time.

[0082] This technical solution, by introducing personalized threshold setting technology and real-time data fusion methods, constructs an intervention execution mechanism oriented towards dynamic changes in patient status, effectively improving the accuracy and timeliness of intervention responses. Based on each patient's historical remission level and characteristics of altered consciousness, the system sets a unique intervention trigger threshold, making intervention decisions more aligned with individual physiological characteristics. Simultaneously, real-time collected feedback data is continuously integrated into the current intervention intensity curve, enabling continuous correction and dynamic optimization of stimulus intensity, avoiding the problem of neglecting individual differences in traditional fixed stimulus patterns. This strategy ensures that the intervention output is always within the optimal response range, significantly enhancing the individual adaptability and rhythmic stability of treatment, providing solid technical support for achieving precision medicine and intelligent intervention.

[0083] Using personalized threshold setting technology, the specific formula for adjusting the triggering conditions for interventions in consciousness disorders based on individual differences is as follows: ;in, The threshold representing individual i, This represents the average of the historical data of individual i. This represents the standard deviation of the historical data for individual i. This represents the standard deviation adjustment parameter, and i represents the individual number index.

[0084] This technical solution utilizes personalized threshold setting technology to dynamically adjust the triggering conditions for interventions in cases of altered consciousness based on each individual's historical data characteristics. Specifically, the system first extracts two key statistical parameters from an individual's past intervention response data: the mean and standard deviation, representing the individual's baseline remission level and fluctuation range, respectively. Then, these two values ​​are linearly combined according to a formula to obtain the individual's dynamic intervention threshold. The threshold is calculated as the sum of the individual's historical mean and its standard deviation multiplied by a standard deviation adjustment parameter, where the adjustment parameter controls threshold sensitivity. The system calculates and stores the threshold for each individual based on their ID number. When real-time feedback data falls below this threshold, the intervention mechanism is triggered, thus achieving an intelligent intervention logic that adaptively adjusts according to individual state changes. This solution effectively improves the intervention system's adaptability to patient differences by transforming the intervention triggering conditions from fixed values ​​to personalized thresholds dynamically calculated based on individual historical data. Personalized threshold setting makes the intervention strategy more aligned with each patient's physiological rhythms and symptom changes, avoiding over- or under-intervention and contributing to improved accuracy and safety of the intervention. Meanwhile, by introducing a standard deviation adjustment parameter, the sensitivity of the intervention threshold can be flexibly controlled, enabling the system to maintain a stable response at different disease stages, which significantly enhances the scientific nature, robustness, and clinical adaptability of the intervention logic.

[0085] This represents the intervention trigger threshold for the i-th individual (i.e., a specific patient) in the actual intervention. This value is the core criterion for the system to determine whether to initiate an intervention operation. Setting method: By... and The calculation is performed, but not directly set. The system updates once every 24 hours during execution. . The average of individual i's historical remission feedback data is a central trend indicator reflecting the level of intervention effectiveness for that individual. Setup: The system automatically extracts the consciousness impairment improvement score data collected from the individual within the last 7 days, performs outlier removal and normalization, and then calculates the average of all valid data points. Example: If individual i's 7-day data contains 1008 valid feedback points (once every 10 minutes), then... This is the arithmetic mean of the 1008 standardized scores. Range: generally between 0.3 and 0.8, depending on the individual's symptom type and remission response. This represents the standard deviation of individual i's feedback data, reflecting the degree of symptom fluctuation for that patient. A larger standard deviation indicates greater patient instability, requiring more sensitive intervention strategies; a smaller standard deviation indicates greater stability. Setting method: [Insert setting here here, but the text doesn't need to be translated.] Same source, calculated based on the same dataset. Range: typically between 0.05 and 0.2, the value affects the final sensitivity of Ti. Used to adjust standard deviation The weights in the threshold calculation are the control variables of this scheme. The larger the value, the more likely it is to result in... The more conservative; The smaller the value, the more sensitive the system is to intervention.

[0086] Standard deviation adjustment parameter The specific formula is as follows: ;

[0087] in, This represents the standard deviation adjustment parameter. This represents the variance of the feedback data for individual i. This represents the time delay from the start of intervention to the generation of a response in individual i. This represents the frequency at which individual i receives intervention per unit of time, where i represents the individual's index.

[0088] By introducing the variance of individual feedback data Response time delay Intervention frequency per unit time Based on core indicators, a composite parameter adjustment model is constructed. Specifically, The calculation of the value takes into account three factors: First, The variance reflects the stability of individual feedback data; a larger variance indicates more drastic fluctuations in individual state and higher unpredictability in response to stimuli. Secondly, This reflects the time delay from intervention initiation to response; the longer the delay, the less sensitive the individual is to the intervention, and the need to increase trigger sensitivity. Third... This indicates the frequency with which the individual actually receives intervention per unit of time. If the frequency is too high, the system needs to appropriately reduce it. To avoid over-intervention, the formula structure employs logarithmic functions and fractional terms to achieve non-linear compression and dynamic harmonization of different indicators, thereby automatically outputting the value best suited to the individual's state. The value. This model is invoked by the system in real time and updated periodically, serving as the threshold for subsequent interventions. The precise calculations provide key weight support.

[0089] Time delay from intervention to response in individual i Specifically, it is the difference between the time point when individual i first exhibits a measurable response and the time point when individual i first receives the intervention. By continuously monitoring the physiological or behavioral feedback data of individual i after the intervention, the moment when the response first exceeds the preset relief threshold is identified as the "time point of measurable response"; simultaneously, the system timestamp of the individual's first receipt of this round of intervention is recorded, and the time difference between the two is the "time point of measurable response". The actual value is determined by the system using timestamps with second precision to mark all data. Combined with continuous feedback data such as symptom scores, neurological reflex indicators, and consciousness arousal scores, a window sliding algorithm is used to identify the "first significant response," typically defined as an increase in score exceeding 0.2 as a valid response. This definition method avoids abstract or subjective judgment, minimizing delay. It has clear and stable quantitative basis, which can be directly used in subsequent parameter calculations and intervention strategy adjustments.

[0090] like Figure 2As shown, a method for interactive awakening of consciousness disorders based on traditional Chinese medicine (TCM) rules is also provided for intelligent terminals. The method employs the aforementioned TCM-based intelligent terminal interactive awakening system for consciousness disorders. The method includes: acquiring multidimensional clinical data of patients with consciousness disorders; constructing an initial patient clinical information matrix through data cleaning and standardization to obtain a structured dataset of consciousness disorder features; obtaining a preliminary intervention intensity curve for the structured dataset; using boundary condition constraint technology to set upper and lower limits for intervention intensity and time, and adjusting the importance parameters of time points through a smoothness weight allocation method to obtain a smooth intervention execution plan; collecting real-time feedback data on the improvement of consciousness disorders during the intervention period based on the smooth intervention execution plan; adjusting rule base parameters through dynamic rule update technology if the feedback data deviates from a preset threshold range, and updating the intervention schedule and stimulus intensity curve by combining interpolation error correction and curve continuity verification to determine an optimized intervention execution plan; and using personalized threshold setting technology to adjust the triggering conditions for consciousness awakening intervention according to individual differences, while integrating the latest feedback data into the intervention intensity curve through a real-time data fusion method to obtain the final dynamic intervention execution logic.

[0091] First, the system acquires multidimensional clinical data from patients with altered consciousness using multimodal sensors, including physiological signals and behavioral manifestations. After outlier removal and normalization, a structured dataset of altered consciousness features is generated. Next, the system uses symptom matching rules and state classification methods, combined with individual frequency and duration characteristics, to generate a personalized symptom priority vector. Based on this vector, a timetable generation algorithm initially determines intervention time points, and an intensity decision model calculates an initial intervention intensity curve using parameters such as feedback volatility. Furthermore, boundary conditions are used to set upper and lower limits for time and intensity, and a smoothness weight is introduced to adjust the intervention density at important time points, forming a stable and executable smooth intervention plan. Once the intervention phase begins, the system collects patient feedback in real time. If the score deviates from the set threshold, the system calls a dynamic rule update mechanism to adjust the rule base, while simultaneously performing error interpolation and curve continuity checks, automatically iteratively optimizing the intervention plan. Finally, the system, considering individual patient characteristics, uses a formulaic threshold setting method to adjust intervention trigger conditions and continuously integrates new feedback data into the intervention intensity curve, achieving real-time closed-loop optimization. This results in the output of the final dynamic intervention logic, ensuring that the intervention strategy at each time point is both individually adaptable and physiologically reasonable.

[0092] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A smart terminal interactive awakening system for consciousness disorders based on traditional Chinese medicine principles, characterized in that: The system includes: The data acquisition module obtains multidimensional clinical data of patients with disorders of consciousness. Through data cleaning and standardization, it constructs an initial patient clinical information matrix to obtain a structured dataset of characteristics of disorders of consciousness. The intervention plan generation module generates a preliminary intervention intensity curve based on a structured dataset of consciousness impairment characteristics. Based on this preliminary intervention intensity curve, boundary condition constraint techniques are used to set upper and lower limits for intervention intensity and time. At the same time, the importance parameters of time points are adjusted through a smoothness weight allocation method to obtain a smooth intervention execution plan. The data feedback and analysis module collects real-time feedback data on the improvement of patients' consciousness impairment during the intervention, based on the smooth intervention execution plan. If the feedback data deviates from the preset threshold range, the rule base parameters are adjusted through dynamic rule update technology. At the same time, combined with interpolation error correction and curve continuity verification, the intervention schedule and stimulus intensity curve are updated to determine the optimized intervention execution plan. The personalized adaptation module uses personalized threshold setting technology to adjust the triggering conditions for consciousness awakening intervention based on individual differences for the optimized intervention implementation plan. At the same time, it integrates the latest feedback data into the intervention intensity curve through real-time data fusion method to obtain the final dynamic intervention implementation logic. The specific formula for adjusting the triggering conditions for intervention in consciousness disorders based on individual differences using personalized threshold setting technology is as follows: ； in, The threshold representing individual i, This represents the average of the historical data of individual i. This represents the standard deviation of the historical data for individual i. This represents the standard deviation adjustment parameter, where i represents the individual number index; the standard deviation adjustment parameter The specific formula is as follows: ； in, This represents the standard deviation adjustment parameter. This represents the variance of the feedback data for individual i. This represents the time delay from the start of intervention to the generation of a response in individual i. This represents the frequency at which individual i receives intervention per unit of time, where i represents the individual's index.

2. The intelligent terminal consciousness disorder interactive awakening system based on traditional Chinese medicine rules according to claim 1, characterized in that: The preliminary intervention intensity curve obtained from the structured dataset of characteristics of consciousness disorders includes: For a structured dataset of consciousness disorder features, a pre-established symptom matching rule is used in conjunction with a patient status classification method to classify patient status into acute and remission phases, thereby determining the classification result of the patient's consciousness disorder.

3. The intelligent terminal consciousness disorder interactive awakening system based on traditional Chinese medicine rules according to claim 2, characterized in that: The preliminary intervention intensity curve obtained from the structured dataset of characteristics of consciousness disorders also includes: Individualized features of frequency and duration of consciousness disorder symptoms were extracted from the classification results of consciousness disorder. Intervention priority ranking logic was used to determine the patient's core symptom intervention needs, resulting in a personalized symptom priority vector.

4. The intelligent terminal consciousness disorder interactive awakening system based on traditional Chinese medicine rules according to claim 3, characterized in that: The preliminary intervention intensity curve obtained from the structured dataset of characteristics of consciousness disorders also includes: Based on personalized symptom priority vectors and combined with historical data weight analysis, a preliminary intervention time point distribution logic is constructed through a timetable generation algorithm to determine the initial intervention timetable for consciousness disorders.

5. The intelligent terminal consciousness disorder interactive awakening system based on traditional Chinese medicine rules according to claim 4, characterized in that: The preliminary intervention intensity curve obtained from the structured dataset of characteristics of consciousness disorders also includes: For the initial intervention timetable for consciousness disorders, a time interval subdivision logic was adopted to divide the timetable into smaller time units. At the same time, the peak period of consciousness disorders was identified as the core intervention point through key node extraction technology, resulting in a detailed intervention time framework.

6. The intelligent terminal consciousness disorder interactive awakening system based on traditional Chinese medicine rules according to claim 5, characterized in that: The preliminary intervention intensity curve obtained from the structured dataset of characteristics of consciousness disorders also includes: From the detailed intervention timeframe, combined with the intensity decision model, the initial stimulus intensity is calculated, and the stimulus intensity at the intermediate point between adjacent key nodes is calculated using the intensity value interpolation formula, thus obtaining the preliminary intervention intensity curve.

7. The intelligent terminal consciousness disorder interactive awakening system based on traditional Chinese medicine rules according to claim 1, characterized in that, The time delay from the start of intervention to the generation of a response in individual i Specifically, it is the difference between the time point when individual i first exhibits a measurable response and the time point when individual i first receives intervention.

8. A method for interactive awakening of consciousness disorders on intelligent terminals based on traditional Chinese medicine rules, employing the interactive awakening system for intelligent terminals based on traditional Chinese medicine rules as described in any one of claims 1-7, characterized in that... The method includes: We acquire multidimensional clinical data of patients with disorders of consciousness, and through data cleaning and standardization, we construct an initial patient clinical information matrix to obtain a structured dataset of characteristics of disorders of consciousness. For the structured dataset of consciousness disorder features, a preliminary intervention intensity curve is obtained. For the preliminary intervention intensity curve, boundary condition constraint technology is used to set upper and lower limits for intervention intensity and time. At the same time, the importance parameters of time points are adjusted by the smoothness weight allocation method to obtain a smooth intervention execution plan. Based on the smooth intervention implementation plan, real-time feedback data on the improvement of patients' consciousness impairment during the intervention period is collected. If the feedback data deviates from the preset threshold range, the rule base parameters are adjusted through dynamic rule update technology. At the same time, combined with interpolation error correction and curve continuity verification, the intervention schedule and stimulus intensity curve are updated to determine the optimized intervention implementation plan. For the optimized intervention implementation plan, a personalized threshold setting technology is adopted to adjust the triggering conditions of consciousness arousal intervention according to individual differences. At the same time, the latest feedback data is integrated into the intervention intensity curve through real-time data fusion method to obtain the final dynamic intervention implementation logic. The specific formula for adjusting the triggering conditions for intervention in consciousness disorders based on individual differences using personalized threshold setting technology is as follows: ； in, The threshold representing individual i, This represents the average of the historical data of individual i. This represents the standard deviation of the historical data for individual i. This represents the standard deviation adjustment parameter, and i represents the individual ID index; The standard deviation adjustment parameter The specific formula is as follows: ； in, This represents the standard deviation adjustment parameter. This represents the variance of the feedback data for individual i. This represents the time delay from the start of intervention to the generation of a response in individual i. This represents the frequency at which individual i receives intervention per unit of time, where i represents the individual's index.

Images