Platelet aggregation inhibitor

Abstract

The application provides a highland crowd health state grading evaluation method and system based on a four-dimensional state space. Specifically, the application performs pre-classification on a subject crowd based on multi-dimensional data such as physiological, psychological, cognitive and sleep data of the subject crowd with highland characteristics. A double-layer weighting method is used to calculate the weight of each index in the sub-population and the space macro weight of the four dimensions of physiology, psychology, cognition and sleep. The health state boundary is determined according to the weight, the closeness of the individual score to the health state is evaluated, and the closeness set is clustered in the back end to dynamically generate the health grade classification boundary. Through the health grade classification boundary, the health state grading of the individual is realized. In addition, according to the health state grading of the individual, a corresponding intervention scheme can be provided. The method of the application adopts four-dimensional data, a double-layer weighting architecture and a dynamically generated classification boundary, and can effectively identify individuals in a sub-health state.

Description

Technical Field

[0001] This invention relates to the field of health assessment technology, specifically a method for classifying and assessing the health status of high-altitude populations based on a four-dimensional state space. It is particularly suitable for the quantitative identification and graded early warning of non-organic functional decline and sub-health status during high-altitude exposure. Background Technology

[0002] With increased investment in high-altitude development, a large number of people are moving from low-altitude areas to high-altitude regions. The low-oxygen environment of high altitudes not only induces organic diseases such as acute mountain sickness (AMS), but also leads to "non-organic functional decline" symptoms such as memory loss, slowed reaction time, sleep disorders, and anxiety. For people living in high-altitude areas, this "sub-health state," which has not yet met the criteria for a clinical diagnosis but whose physical functions have already significantly declined, is often more insidious and dangerous than overt diseases, easily leading to long-term physical and mental damage. Accurately identifying this transitional state between disease and health is a pressing issue that needs to be addressed in current high-altitude health protection efforts.

[0003] Currently, assessments of health status at high altitudes primarily focus on two areas: "physiological parameter monitoring and alarm" and "equipment regulation." Existing research collects basic physiological indicators such as blood oxygen saturation and blood pressure, comparing them to preset fixed thresholds. This threshold-based early warning method, based on a single physiological dimension, struggles to identify situations where physiological indicators remain within safe limits, but functional decline has already occurred. Another study constructed a portable long-term health monitoring and adaptive oxygen supply system for high-altitude environments. By incorporating environmental parameters, blood oxygen levels, and GPS information, it monitors and dynamically adjusts the equipment's oxygen output parameters. This method, emphasizing equipment control rather than health assessment, fails to quantify and classify an individual's overall health status and lacks in-depth attention to the complexities of human health, particularly the psychological and cognitive aspects.

[0004] While the aforementioned technologies provide a reference for the prevention and acclimatization assessment of altitude sickness, they still have the following limitations in assessing the sub-health status of people living at high altitudes: First, existing technologies require fixed, pre-defined grading boundaries for classification. This static grading method, which measures dynamic changes, cannot accurately define the individual's state in the early stages of functional disorder or when indicators are in a gray area, and it is difficult to fully reflect the complex impact of the high-altitude environment on the human body. Second, they ignore the "masking effect" of physiological compensation on latent cognitive impairment: the human body may restore routine physiological indicators (such as SpO2) to the normal range in the short term through compensation by the respiratory and circulatory systems (such as increased heart rate and increased red blood cell count). However, due to sleep disorders and nerve fatigue caused by persistent hypoxia, cognitive functions (such as memory and complex decision-making ability) may still be impaired. Existing technologies cannot identify such situations where indicators are normal but physical and mental discomfort persists.

[0005] Therefore, there is an urgent need to develop a method for assessing the physical and mental health of plateau populations that integrates multidimensional data such as physiological, psychological, and cognitive factors, especially a new method for accurately classifying and assessing health, sub-health, and disease states. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for classifying and assessing the health status of high-altitude populations based on a four-dimensional state space obtained from multidimensional data.

[0007] A first aspect of the present invention provides a method for constructing a health status classification model for a high-altitude population, the method comprising the steps of: (S1) Provide a multidimensional dataset of a high-altitude population; the multidimensional dataset includes physiological function data, psychological data, cognitive efficacy data, sleep health data, and environmental exposure characteristic data; (S2) Preprocess the multidimensional dataset to obtain a preprocessed multidimensional dataset; (S3) Using a clustering algorithm and by iteratively minimizing the objective function, the plateau population is classified according to the feature vectors of the environmental exposure feature data in the preprocessed multidimensional dataset, thereby dividing the plateau population into multiple subgroups; (S4) In each subgroup, the micro weight of each indicator in the preprocessed multidimensional dataset is determined using the objective weighting method; (S5) Based on the standardized values ​​of each indicator in the preprocessed multidimensional dataset and the micro-weights, determine the comprehensive score of each subject in the plateau population on four dimensions, and then use the objective weighting method to determine the macro-spatial weight of each dimension; wherein, the four dimensions are physiological dimension, psychological dimension, cognitive dimension and sleep dimension; (S6) Determine the state vector of each subject in the plateau population and the extreme state vector of the plateau population based on the comprehensive score and the macroscopic spatial weight; evaluate the distance between the state vector of each subject and the extreme state vector using a distance algorithm, thereby assessing the health closeness of each subject and obtaining a closeness set; classify the closeness set using a clustering algorithm and by iteratively minimizing the objective function; determine the evaluation criteria of the plateau population based on the cluster center of each cluster, thereby obtaining a plateau population health status classification model.

[0008] In another preferred embodiment, the high-altitude population includes individuals with long-term and short-to-medium-term exposure to the high-altitude environment.

[0009] In another preferred embodiment, the physiological function data includes: respiratory system data, nervous system data, circulatory system data, and digestive and metabolic data.

[0010] In another preferred embodiment, the respiratory system data includes blood oxygen saturation, vital capacity, and respiratory rate.

[0011] In another preferred embodiment, the nervous system data includes resting heart rate and multidimensional heart rate variability (HRV) indicators.

[0012] In another preferred embodiment, the multidimensional heart rate variability index includes: time-domain index, frequency-domain index, and nonlinear kinetic index.

[0013] In another preferred embodiment, the time-domain indices include the standard deviation of normal heart rate intervals (SDNN) and the root mean of the squared differences (RMSSD).

[0014] In another preferred embodiment, the frequency domain parameters include: Total Power, Low Frequency (LF), High Frequency (HF), and Low Frequency / High Frequency Ratio (LF / HF).

[0015] In another preferred embodiment, the nonlinear dynamics metrics include: sample entropy, Poincaré dimension, approximate entropy, detrended volatility analysis (DFA), and correlation dimension (CD).

[0016] In another preferred embodiment, the circulatory system data includes: blood pressure, hemoglobin concentration, hematocrit (HCt), red blood cell count, serum uric acid (UA), and other common blood biochemical indicators.

[0017] In another preferred embodiment, the digestive and metabolic data include body mass index (BMI) and weight maintenance level.

[0018] In another preferred embodiment, the psychological data includes anxiety scores and depression scores.

[0019] In another preferred embodiment, the anxiety score is obtained using the Generalized Anxiety Disorder Scale (GAD-7).

[0020] In another preferred embodiment, the depression score is obtained using the Depression Screening Scale (PHQ-9).

[0021] In another preferred embodiment, the cognitive efficacy data includes a cognitive error frequency score.

[0022] In another preferred embodiment, the cognitive error frequency score is obtained through a cognitive failure questionnaire (CFQ).

[0023] In another preferred embodiment, the sleep health data includes: objective physiological sleep indicators, subjective assessment indicators, and rhythm indicators.

[0024] In another preferred embodiment, the environmental exposure characteristic data includes natural environmental exposure parameters and daily activity and rest load parameters.

[0025] In another preferred embodiment, the natural environment exposure parameters include elevation gain and time spent living in the high-altitude region.

[0026] In another preferred embodiment, the daily activity and work-rest load parameters include: daily work / study time and monthly frequency of irregular work-rest schedules.

[0027] In another preferred embodiment, step (S2) specifically includes: standardizing the data in the multidimensional dataset using the range standardization method; and / or converting the data into a feature vector.

[0028] In another preferred embodiment, the positive and negative data in the multidimensional dataset are standardized using the range standardization method.

[0029] In another preferred embodiment, the calculation method for standardizing the positive data in the multidimensional dataset using the range standardization method is as follows: ,in, The first among the plateau population The first subject was on the The original observations on each indicator; Its dimensionless value after standardization; and The high-altitude populations were respectively in the following period The maximum and minimum values ​​of each indicator.

[0030] In another preferred embodiment, the calculation method for standardizing the negative data in the multidimensional dataset using the range standardization method is as follows: ,in, The first among the plateau population The first subject was on the The original observations on each indicator; For the first The first subject was on the The dimensionless values ​​after standardization of each indicator; and The high-altitude populations were respectively in the following period The maximum and minimum values ​​of each indicator.

[0031] In another preferred embodiment, the The range of values ​​is .

[0032] In another preferred embodiment, the environmental exposure feature data is converted into a feature vector.

[0033] In another preferred embodiment, the feature vector of the environmental exposure feature data is: .

[0034] In another preferred embodiment, step (S3) specifically includes the following steps: (s3.1) Using a clustering algorithm, determine the cluster centers of each subgroup based on the environmental exposure feature data in the preprocessed multidimensional dataset; (s3.2) Based on the cluster centers, the plateau population is classified by iteratively minimizing the objective function, thereby dividing the plateau population into multiple subgroups.

[0035] In another preferred embodiment, in step (s3.1), the clustering algorithm is the K-means++ algorithm.

[0036] In another preferred embodiment, in step (s3.2), the objective function is: ,in, The objective function is... For the divided first Subgroup; For the first A feature vector of a subject's environmental exposure characteristics data; For the first Cluster centers of individual subgroups.

[0037] In another preferred embodiment, the number of subgroups .

[0038] In another preferred embodiment, the subgroup includes an acute exposure group, a chronic cumulative high-load group, and a conventional steady-state adaptation group.

[0039] In another preferred embodiment, in step (S4), the objective weighting method is the CRITIC weighting method.

[0040] In another preferred embodiment, step (S4) specifically includes the step of: in each subgroup, (s4.1) Evaluate the contrast strength of each indicator in the preprocessed multidimensional dataset and assess the conflict of each indicator based on the correlation coefficient between the indicators; (s4.2) Determine the micro weight of each of the indicators based on the contrast intensity and the conflict.

[0041] In another preferred embodiment, the contrast intensity is the standard deviation.

[0042] In another preferred embodiment, the intensity of the contrast is calculated as follows: , in, For the first The comparative strength of each indicator; For the first The first subject was on the The dimensionless values ​​after standardization of each indicator; : No. within subgroup The average of the indicators; The total number of subjects.

[0043] In another preferred embodiment, the conflict is calculated as follows: ,in, For the first Quantitative values ​​of conflict for each indicator; As an indicator With indicators The Pearson correlation coefficient between them; This represents the total number of indicators.

[0044] In another preferred embodiment, the micro-weights are calculated as follows: ,in, For the first The micro-weight of each indicator; For the first The comparative strength of each indicator; For the first Quantitative values ​​of conflict for each indicator; This represents the total number of indicators.

[0045] In another preferred embodiment, in step (S5), the objective weighting method is the entropy weighting method.

[0046] In another preferred embodiment, step (S5) specifically includes the following steps: (s5.1) Based on the standardized values ​​of each indicator in the preprocessed multidimensional dataset and the micro-weights, evaluate the comprehensive score of each subject in the plateau population on the four dimensions; (s5.2) Normalize the composite score of each subject in the subgroup to which the subject belongs to obtain the score weight of each subject in each dimension; (s5.3) Using the information entropy formula, evaluate the information entropy of each dimension according to the score weighting; (s5.4) Determine the macroscopic spatial weight of each dimension based on the information entropy.

[0047] In another preferred embodiment, the comprehensive score is calculated as follows: ,in, For the first The first subject was on the The overall score across all dimensions; For the first The first subject was on the The dimensionless values ​​after standardization of each indicator; For the first The micro-weight of each indicator; .

[0048] In another preferred embodiment, the weighting of the score is calculated as follows: ,in, For the first The first subject was on the The weighting of scores in each dimension; For the first The first subject was on the The overall score across all dimensions; Total number of subjects; .

[0049] In another preferred embodiment, the information entropy is calculated as follows: ,in, For the first Information entropy in one dimension; for The first subject was on the The weighting of scores in each dimension; Total number of subjects; .

[0050] In another preferred embodiment, the macroscopic spatial weight is calculated as follows: ,in, For the first Macro-spatial weights in each dimension; For the first Information entropy in one dimension; .

[0051] In another preferred embodiment, step (S6) specifically includes the following steps: (s6.1) Construct the state vector of each subject in the plateau population based on the comprehensive score and the macro-spatial weight; (s6.2) Determine the extreme state vector of the plateau population based on the extreme values ​​of the state vectors of all subjects in each subgroup; (s6.3) Evaluate the distance between the state vector of each subject and the extreme state vector using a distance algorithm; evaluate the overall health closeness of each subject based on the distance, and obtain a closeness set; (s6.4) Using a clustering algorithm, determine the initial cluster centers of each cluster based on the proximity set; (s6.5) Based on the initial cluster centers, the proximity set is divided into multiple clusters, and new cluster centers are determined by iteratively minimizing the objective function; (s6.6) Determine the evaluation criteria for the plateau population based on the new cluster centers of each cluster, thereby obtaining a health status classification model for the plateau population.

[0052] In another preferred embodiment, the extreme state vector includes a best state vector and a worst state vector.

[0053] In another preferred embodiment, in step (s6.1), the weighted score for each dimension is obtained by multiplying the comprehensive score and the macroscopic spatial weights item by item; a weighted feature matrix is ​​constructed based on the weighted scores of all dimensions, thereby determining the state vector of each subject in the plateau population.

[0054] In another preferred embodiment, step (s6.2) includes: determining the optimal state vector of the plateau population based on the maximum value of the state vector of all subjects in each subgroup; and determining the minimum state vector of the plateau population based on the minimum value of the state vector of all subjects in each subgroup.

[0055] In another preferred embodiment, in step (s6.3), the distance algorithm includes Euclidean distance.

[0056] In another preferred embodiment, step (s6.3) includes: determining the distance between the state vector of each subject and the optimal state vector and the worst state vector using Euclidean distance; and evaluating the overall health closeness of each subject based on the distance to obtain a closeness set.

[0057] In another preferred embodiment, the distance between the state vector of each subject and the optimal state vector is calculated using Euclidean distance as follows: ,in, The distance to the optimal state vector; For the first The first subject was on the A state vector in each dimension; For the first An optimal state vector in each dimension.

[0058] In another preferred embodiment, the distance between the state vector and the range state vector of each subject is calculated using Euclidean distance as follows: ,in, The distance to the range state vector; For the first The first subject was on the A state vector in each dimension; For the first A range state vector with 1 dimension.

[0059] In another preferred embodiment, the proximity is calculated as follows: ,in, For the first The closeness of each subject; The distance to the optimal state vector; This is the distance to the range state vector.

[0060] In another preferred embodiment, in step (s6.4), the clustering algorithm is the K-means++ algorithm.

[0061] In another preferred embodiment, step (s6.4) includes the following steps: (s6.4.1) Using the K-means++ algorithm, a sample point is randomly selected from the proximity set as the first initial cluster center; (s6.4.2) Determine the distance between the remaining sample points and the first initial cluster center using a distance algorithm, and evaluate the probability that each remaining sample point will be selected as the next cluster center; (s6.4.3) Select the next cluster center using a roulette wheel method based on the aforementioned probability; (s6.4.4) Repeat step (s6.4.3) until all selections are complete. The initial cluster centers.

[0062] In another preferred embodiment, in step (s6.4.2), the probability is calculated as follows: ,in, This represents the probability that a sample point will be selected as the next cluster center. The distance between the sample point and the first initial cluster center; In another preferred embodiment, in step (s6.4.2), the distance algorithm includes Euclidean distance.

[0063] In another preferred embodiment, in step (s6.4.4), the number of clusters... .

[0064] In another preferred embodiment, step (s6.5) includes: determining the remaining sample points and the distance obtained in step (s6.4) using a distance algorithm. The initial cluster centers are used to divide each sample point into the cluster with the smallest initial cluster center; the initial cluster centers of each cluster are updated, and new cluster centers are determined by iteratively minimizing the objective function.

[0065] In another preferred embodiment, the initial cluster centers of each cluster are calculated as follows: ,in, For the first One cluster; For the first Sample size within each cluster; For the first The set of proximity scores for all sample points within a cluster; For the first Within the cluster, the _ The closeness of each sample point.

[0066] In another preferred embodiment, in step (s6.5), the objective function is: ,in, The objective function is... For the first The set of proximity scores for all sample points within a cluster; For the first Within the cluster, the _ The closeness of each sample point; For the first The updated cluster centers in each cluster.

[0067] In another preferred embodiment, step (s6.6) includes: dividing the clusters into "pathological risk" clusters based on the numerical value of the new cluster centers for each cluster. "Sub-health functional decline" cluster center and "healthy homeostasis" cluster center Based on the cluster centers, pathological risk boundary thresholds and health homeostasis boundary thresholds are identified, and a dynamic evaluation standard sequence with the pathological risk boundary thresholds and health homeostasis boundary thresholds as boundaries is formed, thereby obtaining the health status classification model of the plateau population.

[0068] In another preferred embodiment, in step (s6.6), the pathological risk boundary threshold is calculated as follows: ,in, This is the threshold for pathological risk. It serves as the cluster center for "pathological risk"; It serves as the cluster center for "sub-health functional decline".

[0069] In another preferred embodiment, in step (s6.6), the healthy steady-state boundary threshold is calculated as follows: ,in, This represents the healthy steady-state boundary threshold. It serves as the cluster center for "sub-health functional decline"; It serves as a cluster center for "healthy homeostasis".

[0070] In another preferred embodiment, the dynamic evaluation criterion sequence includes ,in, This is the threshold for pathological risk. This represents the healthy steady-state boundary threshold.

[0071] In another preferred embodiment, the method further includes the step of: (S7) Monitor the cumulative sample size in the multidimensional dataset in real time; when the cumulative sample size meets the preset triggering mechanism, repeat steps (S2) to (S6) to update the evaluation criteria for the plateau population, thereby obtaining an updated health status classification model for the plateau population.

[0072] In another preferred embodiment, the preset triggering mechanism includes: the current cumulative sample size reaches an integer multiple of the initial baseline sample size.

[0073] A second aspect of the present invention provides a system for classifying the health status of people living in high-altitude areas, the system comprising: (M1) Input module, the input module is configured to input data, the data including a multidimensional dataset of the test object, the multidimensional dataset including physiological function data, psychological data, cognitive efficacy data, sleep health data and environmental exposure characteristic data; (M2) Analysis module, configured to analyze the data and obtain analysis results; the analysis module includes: (m2.1) A grading unit, wherein the grading unit is configured as a plateau population health status grading model, wherein the plateau population health status grading model grades the test object according to the multidimensional dataset to obtain grading results; the plateau population health status grading model is constructed using the method described in the first aspect of the present invention; (M3) Output module, which is configured to output the results of the analysis module.

[0074] In another preferred embodiment, the analysis module further includes: (m2.2) Intervention unit, the intervention unit is configured to perform the following operations: based on the grading results, if the subject to be tested is at pathological risk or experiencing functional decline in sub-health, then determine the dimension with the lowest comprehensive score among the subjects to be tested, thereby determining the corresponding intervention strategy; Specifically, if the dimension with the lowest overall score is physiological function, an intervention strategy of "proactive medical repair" is provided; if the dimension with the lowest overall score is psychological and / or cognitive efficacy, an intervention strategy of "psychological adjustment" is provided; and if the dimension with the lowest overall score is sleep health, an intervention strategy of "sleep intervention" is provided.

[0075] In another preferred embodiment, the intervention strategy of the "active medical repair" includes: increasing the frequency of oxygen inhalation, mandatory rest, and mandatory reduction of physical activity load.

[0076] In another preferred embodiment, the intervention strategies for “psychological adjustment” include: mindfulness relaxation training and suspending high cognitive load activities.

[0077] In another preferred embodiment, the intervention strategy of the "sleep intervention" includes: increasing the ventilation rate of the sleep environment, reducing the carbon dioxide concentration of the sleep environment, increasing the light blocking of the sleep environment, controlling the noise of the sleep environment, and supplementing oxygen at night.

[0078] In another preferred embodiment, the (M2) analysis module further includes: (m2.3) Update unit, which is configured to monitor the data scale input by the (M1) input module, and when a preset triggering mechanism is met, execute steps (S2)-(S6) in the method of the first aspect of the present invention to update the evaluation criteria of the plateau population and obtain an updated plateau population health status classification model, so that the (m2.1) classification unit can be configured as the updated plateau population health status classification model.

[0079] A third aspect of the present invention provides an electronic device, including a processor and a memory, the memory having a plurality of executable instructions, the processor being configured to read the instructions and execute steps in a method for grading and assessing the health status of a test object; the method comprising the steps of: (1) Provide a multidimensional dataset of a test subject, the multidimensional dataset including physiological function data, psychological data, cognitive efficacy data, sleep health data and environmental exposure characteristic data; (2) Preprocess the multidimensional dataset to obtain a preprocessed multidimensional dataset; (3) Based on the feature vectors of the environmental exposure feature data in the preprocessed multidimensional dataset, the test object is matched to a predetermined exposure feature subgroup; based on the predetermined micro-weights in the exposure feature subgroup to which the test object belongs and the standardized values ​​of each indicator in the preprocessed multidimensional dataset, the comprehensive score of the test object is determined; based on the comprehensive score and the predetermined macro-spatial weights, the state vector of the test object is determined; the distance between the state vector of the test object and the predetermined extreme state vector is evaluated by a distance algorithm, thereby evaluating the comprehensive health proximity of the test object; (4) Input the comprehensive health closeness of the test subject into the plateau population health status classification model and compare it with the predetermined boundary threshold therein, so as to classify the test subject; The boundary thresholds include pathological risk boundary thresholds. and healthy steady-state boundary threshold If the overall health relevance of the tested object is... If the overall health closeness of the tested object is good, it indicates that the object is in a healthy homeostasis. If the test subject is at pathological risk, then the overall health proximity C of the test subject satisfies the following conditions. If so, it indicates that the subject is in a state of sub-health and functional decline. The health status classification model for the plateau population is constructed using the method described in the first aspect of this invention.

[0080] In another preferred embodiment, the exposure characteristic subgroup is an acute exposure group, a chronic cumulative high load group, and a conventional steady-state adaptation group.

[0081] In another preferred embodiment, the macroscopic spatial weights are physiological dimension weights, psychological dimension weights, cognitive dimension weights, and sleep dimension weights.

[0082] In another preferred embodiment, the distance algorithm includes Euclidean distance.

[0083] In another preferred embodiment, the extreme state vector is a best state vector and a worst state vector.

[0084] In another preferred embodiment, in step (4), the health status grading model for high-altitude populations further includes providing intervention strategies for the subjects to be tested based on the grading results; Specifically, if the dimension with the lowest overall score is physiological function, an intervention strategy of "proactive medical repair" is provided; if the dimension with the lowest overall score is psychological and / or cognitive efficacy, an intervention strategy of "psychological adjustment" is provided; and if the dimension with the lowest overall score is sleep health, an intervention strategy of "sleep intervention" is provided.

[0085] A fourth aspect of the present invention provides a computer-readable storage medium storing computer-executable instructions, which, when read and executed by a processor, implement steps in a method for grading and assessing the health status of a test object; the method includes the following steps: (1) Provide a multidimensional dataset of a test subject, the multidimensional dataset including physiological function data, psychological data, cognitive efficacy data, sleep health data and environmental exposure characteristic data; (2) Preprocess the multidimensional dataset to obtain a preprocessed multidimensional dataset; (3) Based on the feature vectors of the environmental exposure feature data in the preprocessed multidimensional dataset, the test object is matched to a predetermined exposure feature subgroup; based on the predetermined micro-weights in the exposure feature subgroup to which the test object belongs and the standardized values ​​of each indicator in the preprocessed multidimensional dataset, the comprehensive score of the test object is determined; based on the comprehensive score and the predetermined macro-spatial weights, the state vector of the test object is determined; the distance between the state vector of the test object and the predetermined extreme state vector is evaluated by a distance algorithm, thereby evaluating the comprehensive health proximity of the test object; (4) Input the comprehensive health closeness of the test subject into the plateau population health status classification model and compare it with the predetermined boundary threshold therein, so as to classify the test subject; The boundary thresholds include pathological risk boundary thresholds. and healthy steady-state boundary threshold If the overall health relevance of the tested object is... If the overall health closeness of the tested object is good, it indicates that the object is in a healthy homeostasis. If the test subject is at pathological risk, then the overall health proximity C of the test subject satisfies the following conditions. If so, it indicates that the subject is in a state of sub-health and functional decline. The health status classification model for the plateau population is constructed using the method described in the first aspect of this invention.

[0086] In another preferred embodiment, the exposure characteristic subgroup is an acute exposure group, a chronic cumulative high load group, and a conventional steady-state adaptation group.

[0087] In another preferred embodiment, the macroscopic spatial weights are physiological dimension weights, psychological dimension weights, cognitive dimension weights, and sleep dimension weights.

[0088] In another preferred embodiment, the distance algorithm includes Euclidean distance.

[0089] In another preferred embodiment, the extreme state vector is a best state vector and a worst state vector.

[0090] In another preferred embodiment, in step (4), the health status grading model for high-altitude populations further includes providing intervention strategies for the subjects to be tested based on the grading results; Specifically, if the dimension with the lowest overall score is physiological function, an intervention strategy of "proactive medical repair" is provided; if the dimension with the lowest overall score is psychological and / or cognitive efficacy, an intervention strategy of "psychological adjustment" is provided; and if the dimension with the lowest overall score is sleep health, an intervention strategy of "sleep intervention" is provided.

[0091] A fifth aspect of the present invention provides a computer program product comprising computer-executable instructions, which, when executed by a processor, implement steps in a method for grading and assessing the health status of a test object; the method comprising the steps of: (1) Provide a multidimensional dataset of a test subject, the multidimensional dataset including physiological function data, psychological data, cognitive efficacy data, sleep health data and environmental exposure characteristic data; (2) Preprocess the multidimensional dataset to obtain a preprocessed multidimensional dataset; (3) Based on the feature vectors of the environmental exposure feature data in the preprocessed multidimensional dataset, the test object is matched to a predetermined exposure feature subgroup; based on the predetermined micro-weights in the exposure feature subgroup to which the test object belongs and the standardized values ​​of each indicator in the preprocessed multidimensional dataset, the comprehensive score of the test object is determined; based on the comprehensive score and the predetermined macro-spatial weights, the state vector of the test object is determined; the distance between the state vector of the test object and the predetermined extreme state vector is evaluated by a distance algorithm, thereby evaluating the comprehensive health proximity of the test object; (4) Input the comprehensive health closeness of the test subject into the plateau population health status classification model and compare it with the predetermined boundary threshold therein, so as to classify the test subject; The boundary thresholds include pathological risk boundary thresholds. and healthy steady-state boundary threshold If the overall health relevance of the tested object is... If the overall health closeness of the tested object is good, it indicates that the object is in a healthy homeostasis. If the test subject is at pathological risk, then the overall health proximity C of the test subject satisfies the following conditions. If so, it indicates that the subject is in a state of sub-health and functional decline. The health status classification model for the plateau population is constructed using the method described in the first aspect of this invention.

[0092] In another preferred embodiment, the exposure characteristic subgroup is an acute exposure group, a chronic cumulative high load group, and a conventional steady-state adaptation group.

[0093] In another preferred embodiment, the macroscopic spatial weights are physiological dimension weights, psychological dimension weights, cognitive dimension weights, and sleep dimension weights.

[0094] In another preferred embodiment, the distance algorithm includes Euclidean distance.

[0095] In another preferred embodiment, the extreme state vector is a best state vector and a worst state vector.

[0096] In another preferred embodiment, in step (4), the health status grading model for high-altitude populations further includes providing intervention strategies for the subjects to be tested based on the grading results; Specifically, if the dimension with the lowest overall score is physiological function, an intervention strategy of "proactive medical repair" is provided; if the dimension with the lowest overall score is psychological and / or cognitive efficacy, an intervention strategy of "psychological adjustment" is provided; and if the dimension with the lowest overall score is sleep health, an intervention strategy of "sleep intervention" is provided.

[0097] It should be understood that, within the scope of this invention, the above-described technical features of this invention and the technical features specifically described below (such as in the embodiments) can be combined with each other to form new or preferred technical solutions. Due to space limitations, they will not be described in detail here. Attached Figure Description

[0098] Figure 1 The flowchart of the invention, which is based on a four-dimensional functional state space, is shown.

[0099] Figure 2 The diagram shows the spatial distribution of four-dimensional functional state hierarchy based on K-means++ clustering, which intuitively demonstrates the distribution boundaries of the three state clusters: "healthy homeostasis", "sub-healthy functional decline" and "pathological risk".

[0100] Figure 3 The study displays a four-dimensional health assessment radar chart of the subjects, demonstrating the physiological compensation masking effect and the adaptive judgment boundary.

[0101] Figure 4 This is a schematic diagram of the ROC curve of a plateau personnel physical and mental health evaluation model provided in an embodiment of the present invention.

[0102] Figure 5 This is a schematic diagram of the confusion matrix of a plateau personnel physical and mental health evaluation model provided in an embodiment of the present invention. Detailed Implementation

[0103] Through extensive and in-depth research, the inventors have, for the first time, constructed a method and system for classifying and assessing the health status of a high-altitude population based on multidimensional data encompassing physiological, psychological, cognitive, and sleep health aspects. Specifically, firstly, this invention collects multidimensional data from the high-altitude population and performs preliminary triage based on environmental exposure characteristics within this data, dividing the population into three subgroups. Secondly, using the "CRITIC-EWM" two-layer weighting method, the micro-weights of each indicator are calculated within each subgroup, followed by the spatial macro-weights of the four dimensions of physiological, psychological, cognitive, and sleep health. Based on these weights, the health status boundary is determined, the closeness between individual scores and health status is assessed, and the closeness set is clustered in the backend to dynamically generate a health level classification boundary. This boundary is then used to classify the health status of individual individuals and provide corresponding intervention suggestions. Furthermore, this assessment process possesses adaptive iterative capabilities, automatically triggering recalibration and optimization of the evaluation standard sequence as the accumulated amount of subject data increases. Based on these findings, this invention was completed.

[0104] Furthermore, the present invention provides a method for classifying and assessing the health status of high-altitude populations based on a four-dimensional functional state space, comprising: S1: Provides multidimensional heterogeneous data. The multidimensional dataset includes physiological function data, psychological data, cognitive efficacy data, sleep health data, and environmental exposure characteristic data.

[0105] S2: Preprocess the multidimensional heterogeneous data to obtain a preprocessed multidimensional dataset.

[0106] S3: Using a clustering algorithm and iteratively minimizing the objective function, the plateau population is classified according to the feature vectors of environmental exposure feature data in the preprocessed multidimensional dataset, thereby dividing the plateau population into multiple subgroups; the aim is to eliminate subjective prior errors and achieve pre-diversion of homogeneous exposure features.

[0107] S4: In each subgroup, the micro-weight of each indicator in the preprocessed multidimensional dataset is determined using an objective weighting method; that is, weighting is performed within the micro-dimension by utilizing the contrast strength and conflict between indicators.

[0108] S5: Determine the comprehensive score of each subject in the plateau population on four dimensions based on the standardized values ​​of each indicator in the preprocessed multidimensional dataset and the micro-weights, and then use the objective weighting method to determine the macro-spatial weight of each dimension; wherein, the four dimensions are physiological dimension, psychological dimension, cognitive dimension and sleep dimension.

[0109] S6: Determine the state vector of each subject in the plateau population and the extreme state vector of the plateau population based on the comprehensive score and the macroscopic spatial weight; evaluate the distance between the state vector of each subject and the extreme state vector using a distance algorithm, thereby assessing the health closeness of each subject and obtaining a closeness set; classify the closeness set using a clustering algorithm and by iteratively minimizing the objective function; determine the evaluation criteria of the plateau population based on the cluster center of each cluster, thereby obtaining a plateau population health status grading model.

[0110] S7: Monitor the cumulative sample size in the multidimensional dataset in real time; when the cumulative sample size meets the preset triggering mechanism, re-execute steps (S2) to (S6) to update the evaluation criteria for the plateau population, thereby obtaining an updated health status classification model for the plateau population and realizing the adaptive evolution of the evaluation criteria.

[0111] After obtaining the health status grading model, in a preferred embodiment of practical application, this method may further include differentiated proactive health interventions based on the grading results: Providing differentiated proactive health intervention strategies. If the subject is at risk of pathological disease or experiencing a decline in sub-health function, the dimension with the lowest overall score is identified, and corresponding "proactive medical repair," "psychological adjustment," or "sleep intervention" repair and regulation strategies are provided. If the disadvantage is in a physiological dimension, suggestions are provided to increase the frequency of oxygen inhalation, force rest, or forcefully reduce physical activity load; if the disadvantage is in a psychological or cognitive dimension, suggestions are provided to mindfulness relaxation training or postpone high cognitive load activities; if the disadvantage is in a sleep health dimension, suggestions are provided to improve the ventilation rate of the sleep environment, reduce the carbon dioxide concentration of the sleep environment, increase the light blocking of the sleep environment, control the noise of the sleep environment, or provide nighttime oxygen supplementation.

[0112] Preferably, the physiological function dimension feature data mentioned in step S1 includes, but is not limited to: Respiratory system parameters: blood oxygen saturation, vital capacity, and respiratory rate; Neurological parameters: resting heart rate and multidimensional heart rate variability (HRV); Circulatory system parameters: blood pressure, hemoglobin concentration, hematocrit (HCt), red blood cell count, serum uric acid (UA), and other common blood biochemical indicators; Digestive and metabolic parameters: Body mass index (BMI) and weight maintenance level.

[0113] More preferably, the multidimensional heart rate variability index includes: Time-domain metrics: SDNN, RMSSD; Frequency domain metrics: Total Power, Low Frequency (LF), High Frequency (HF), Low Frequency / High Frequency Ratio (LF / HF); Nonlinear dynamics metrics: sample entropy, Poincaré dimension, approximate entropy, detrended fluctuation analysis (DFA), and correlation dimension (CD).

[0114] Preferably, the psychological and emotional dimension feature data mentioned in step S1 includes: anxiety score obtained based on the Generalized Anxiety Disorder Scale (GAD-7) and depression score obtained based on the Depression Screening Scale (PHQ-9).

[0115] Preferably, the cognitive efficacy dimension feature data mentioned in step S1 includes: a score obtained from the Cognitive Failure Questionnaire (CFQ) to quantify the frequency of daily cognitive errors of the subjects.

[0116] Preferably, the sleep health dimension feature data mentioned in step S1 includes: Objective physiological sleep indicators: the proportion of deep sleep, REM sleep, light sleep, number of awakenings at night, total awake time, and time in bed and total sleep time, all collected using wearable devices; Subjective assessment indicators: Based on the Pittsburgh Sleep Quality Index (PSQI) scale, the scores of seven components are: subjective sleep quality, time to fall asleep, sleep duration, sleep efficiency, sleep disorders, use of hypnotics, and daytime dysfunction. Circadian rhythm index: The degree of circadian rhythm disorder calculated based on the absolute difference between the midpoint of sleep on weekdays and the midpoint of sleep on rest days.

[0117] Preferably, the baseline characteristics of environmental and individual behavioral exposures mentioned in step S1 include: Natural environmental exposure parameters: based on the elevation difference calculated from the altitude of the place of residence before entering the plateau and the time spent living in the plateau; Daily activity and work-rest load parameters: daily work / study time and monthly frequency of irregular work-rest schedules.

[0118] Preferably, in step (S2), the preprocessing of the multidimensional dataset specifically includes: standardizing the positive and negative data in the multidimensional dataset using the range standardization method to eliminate dimensional differences; and converting the environmental exposure feature data into feature vectors as input features for subsequent clustering.

[0119] Preferably, the adaptive triage in step S3 specifically includes: using the altitude difference, time spent living in high-altitude areas, daily work / study time, and monthly frequency of irregular work and rest as clustering input features obtained in step S1, the K-means++ algorithm is used to divide the population into acute exposure group, chronic cumulative high load group, and conventional steady-state adaptation group. The algorithm initializes the cluster centers by using distance probability weighting to avoid getting trapped in local optima, aiming to eliminate the prior error caused by the subjective identification of the population by human intervention, and to ensure that the subsequent evaluation is carried out in a homogeneous exposure background.

[0120] Preferably, the objective combination weighting described in steps S4 and S5 specifically refers to: S4: The CRITIC method is used to reflect the contrast intensity by calculating the standard deviation of the indicators and the correlation coefficient between the indicators to reflect the conflict. The aim is to capture the asynchronous change characteristics between physiological indicators and cognitive / psychological indicators and to break the masking effect of physiological compensation on hidden damage. S5: The entropy weight method (EWM) is used to assign weights based on the degree of dispersion of the information in the four dimensions.

[0121] Preferably, the adaptive grading in step S6 specifically includes: calculating the distance between the subject's state vector and the optimal and worst state vectors using Euclidean distance to obtain the closeness; clustering the closeness score set using the K-means++ algorithm, setting the number of clusters k to be an integer greater than or equal to 2; preferably, setting the number of clusters k=3; determining new cluster centers by iteratively minimizing the objective function, and extracting the median value of adjacent cluster centers to confirm the pathological risk boundary threshold T1 and the health homeostasis boundary threshold T2, forming a dynamic evaluation standard sequence.

[0122] Furthermore, the adaptive grading described in step S6 is divided into two stages in practical applications: "benchmark extraction" and "individual mapping." In the benchmark extraction stage, grading boundary thresholds are generated using the initial subject group data. and The system stores the scores of subjects in real time. During the individual mapping stage, the system directly compares the scores of subjects in real time with the stored thresholds to achieve rapid classification and discrimination without calling the clustering engine in real time.

[0123] Preferably, as described in step S7, the present invention further includes a sample-driven iterative update mechanism. The system monitors the cumulative sample size in real time. When the cumulative sample size reaches a preset multiple (e.g., 2 times) of the initial baseline sample size, the system automatically triggers an "evaluation criterion recalibration" command. The clustering iteration process of steps S2 to S6 is re-executed to update the boundary thresholds. and This ensures that the evaluation criteria evolve adaptively.

[0124] Preferably, providing intervention strategies for the test subject based on the grading results specifically includes: the system tracing the lowest-scoring disadvantage dimension; if the disadvantage dimension is sleep health, then suggestions are output to improve the sleep microenvironment, adjust the work and rest schedule, or supplement oxygen at night; if the disadvantage dimension is psychological or cognitive, then suggestions are output to perform mindfulness relaxation training or postpone high cognitive load activities; if the disadvantage dimension is physiological, then suggestions are output to increase the frequency of oxygen inhalation or reduce the physical activity load.

[0125] It should be understood that the specific methods and experimental conditions of the invention described below in varying degrees of detail are intended to provide a substantive understanding of the invention. Definitions of certain terms used in this specification are provided below. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0126] the term As used herein, the terms “containing” or “including (comprise)” can be open-ended, semi-closed, or closed-ended. In other words, the terms also include “consistently made of” or “made of”.

[0127] As used herein, the term “and / or” refers to and covers any and all possible combinations of one or more of the related listed items.

[0128] As used herein, the term "plateau population" is a broad group encompassing various exposure backgrounds and residency characteristics. Due to the complex, cumulative, and heterogeneous effects of plateau environments (such as hypoxia and low pressure) on human physiological functions, psychological emotions, and cognitive efficacy, this group includes not only short- to medium-term exposure personnel who enter plateaus for engineering construction or special missions, but also extensively encompasses long-term migrant residents living in plateau environments, as well as long-term residents in a stable state. The evaluation model constructed in this invention aims to break down the subjective boundaries of population identity through pre-adaptive triage of environmental and daily life exposure characteristics, enabling the model to be generalized and accurately applied to health early warning and proactive intervention for the aforementioned heterogeneous plateau populations. Preferably, the plateau population of this invention includes both long-term and short- to medium-term exposure personnel to the plateau environment.

[0129] The computer system is equipped with at least one processor and a memory. The processor invokes a sequence of computer-executable instructions stored in the memory to implement the evaluation process defined in the claims. Although the flowchart describes the operation steps in a specific logical order, in actual execution, the steps may be processed in parallel, their order adjusted, or partially omitted in some cases. As long as such adjustments do not deviate from the core features of the technical solution described in the claims and achieve the same technical effect, they all fall within the scope of protection of this invention. This flexibility in execution order is determined by the programmable nature of computer instructions.

[0130] Health status classification method for high-altitude populations This invention provides a method for classifying the health status of people living in high-altitude areas, the method comprising the following steps: (1) Provide a multidimensional dataset of a test subject, the multidimensional dataset including physiological function data, psychological data, cognitive efficacy data, sleep health data and environmental exposure characteristic data; (2) Preprocess the multidimensional dataset to obtain a preprocessed multidimensional dataset; (3) Based on the feature vectors of the environmental exposure feature data in the preprocessed multidimensional dataset, the test object is matched to a predetermined exposure feature subgroup; based on the predetermined micro-weights in the exposure feature subgroup to which the test object belongs and the standardized values ​​of each indicator in the preprocessed multidimensional dataset, the comprehensive score of the test object is determined; based on the comprehensive score and the predetermined macro-spatial weights, the state vector of the test object is determined; the distance between the state vector of the test object and the predetermined extreme state vector is evaluated by a distance algorithm, thereby evaluating the comprehensive health proximity of the test object; (4) Input the comprehensive health closeness of the test subject into the plateau population health status classification model and compare it with the predetermined boundary threshold therein, so as to classify the test subject; The boundary thresholds include pathological risk boundary thresholds. and healthy steady-state boundary threshold If the overall health relevance of the tested object is... If the overall health closeness of the tested object is good, it indicates that the object is in a healthy homeostasis. If the test subject is at pathological risk, then the overall health proximity C of the test subject satisfies the following conditions. If so, it indicates that the subject is in a state of sub-health and functional decline.

[0131] Preferably, the exposure characteristic subgroups are acute exposure groups, chronic cumulative high load groups, and conventional steady-state adaptation groups.

[0132] Preferably, the macroscopic spatial weights are physiological dimension weights, psychological dimension weights, cognitive dimension weights, and sleep dimension weights.

[0133] Preferably, the distance algorithm includes Euclidean distance.

[0134] Preferably, the extreme state vector is a best state vector and a worst state vector.

[0135] Preferably, in step (4), the plateau population health status classification model further includes providing intervention strategies for the test subjects based on the classification results; Specifically, if the dimension with the lowest overall score is physiological function, an intervention strategy of "proactive medical repair" is provided; if the dimension with the lowest overall score is psychological and / or cognitive efficacy, an intervention strategy of "psychological adjustment" is provided; and if the dimension with the lowest overall score is sleep health, an intervention strategy of "sleep intervention" is provided.

[0136] The system of the present invention This invention provides a system for classifying the health status of people living in high-altitude areas. The system can be an electronic device, including but not limited to: smartphones, tablets, personal computers, servers, terminals, or other intelligent terminals with data processing capabilities. The system can also be a computer-readable storage medium, such as a hard disk, optical disk, solid-state drive, read-only memory, or flash memory, storing a computer program that, when executed by one or more processors, enables the data processing flow. Furthermore, the system can also be a computer program product containing computer instructions that, when executed by a computer, cause the computer to perform all or part of the steps of the data processing flow.

[0137] The data processing flow can be loaded, deployed, or run on any of the aforementioned electronic devices. Through the collaboration of hardware resources and the software logic defined in the flow, the system can complete specific data acquisition, transmission, calculation, analysis, storage, or presentation tasks.

[0138] The main advantages of this invention include: (1) This invention introduces a multi-dimensional indicator system covering physiological health, mental health, cognitive health and sleep health, and uses the CRITIC algorithm to capture the asynchronous change characteristics between physiological indicators and cognitive indicators. This effectively avoids the false negative results that may be obtained based on a single physiological dimension, and can identify hidden sub-health risks earlier and more accurately.

[0139] (2) The present invention adopts the “CRITIC-EWM” two-layer pure objective combined weighting architecture based on data fluctuation characteristics and information entropy. This architecture is completely driven by real-time data of the test group, effectively avoiding the weight bias caused by subjective experience, and ensuring the objectivity and rigor of the evaluation model in complex medical-engineering cross-situation.

[0140] (3) This invention breaks through the limitations of traditional solutions that only focus on human physiological signs and constructs a four-dimensional assessment system that integrates physiological, psychological, cognitive and sleep health. By deeply integrating physiological signs, psychological emotions, cognitive efficacy and subjective and objective sleep data, it achieves a deep characterization of the entire spectrum of health homeostasis, sub-health risk and disease risk in plateau populations.

[0141] (4) The present invention introduces a mechanism of pre-clustering diversion and back-end clustering to generate hierarchical boundaries. Instead of relying on static global fixed health thresholds, it can perform dynamic hierarchical evaluation of physical and mental health, which improves the robustness of the model in extreme environments.

[0142] (5) This invention not only outputs health rating conclusions, but also provides source-based interventions based on the shortcomings in the four-dimensional scores. For individuals in a sub-healthy state, the system can output differentiated proactive intervention strategies such as "medical repair" (e.g., oxygen inhalation, forced rest), "psychological adjustment" (e.g., mindfulness relaxation training), and "sleep intervention" (e.g., improving environmental ventilation rate, controlling noise) based on the assessment results, transforming risk warnings into actionable closed-loop management recommendations. This mechanism effectively reduces the subjective blind spots and costs of manual screening, and improves the efficiency and operability of large-scale health protection for the plateau population.

[0143] The present invention will be further illustrated below with reference to specific embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. It is also understood that the purpose of describing the present invention in conjunction with the embodiments is to cover other options or modifications that may be derived based on the claims of the present invention. To provide a deep understanding of the invention, many specific details will be included in the following description. The invention may also be practiced without using these details. Furthermore, to avoid confusion or obscuring the focus of the invention, some specific details will be omitted in the description.

[0144] Example 1: A method for classifying and assessing the health status of high-altitude populations based on a four-dimensional functional state space. This embodiment proposes a method for classifying and assessing the health status of high-altitude populations based on a four-dimensional functional state space. The specific steps are as follows: Step S1: Acquisition of multidimensional heterogeneous data 1. Map functional characteristic data to a four-dimensional space of "physiological-psychological-cognitive-sleep": (1) Physiological functional dimensions include, but are not limited to: respiratory system parameters: blood oxygen saturation, vital capacity and respiratory rate; nervous system parameters: resting heart rate and multidimensional heart rate variability (HRV) index; circulatory system parameters: blood pressure, hemoglobin concentration, hematocrit (HCt), red blood cell count, blood uric acid (UA) and other common blood biochemical indicators; digestive and metabolic parameters: body mass index (BMI) and weight maintenance level.

[0145] The multidimensional heart rate variability indicators include: time domain indicators: standard deviation of normal heart rate interval (SDNN), root mean square of continuous differences (RMSSD); frequency domain indicators: total power, low frequency (LF), high frequency (HF), low frequency / high frequency ratio (LF / HF); and nonlinear dynamic indicators: sample entropy, Poincaré dimension, approximate entropy, detrended fluctuation analysis (DFA), and correlation dimension (CD).

[0146] (2) The psychological and emotional dimensions include: anxiety score obtained based on the Generalized Anxiety Disorder Scale (GAD-7) and depression score obtained based on the Depression Screening Scale (PHQ-9).

[0147] (3) The cognitive efficacy dimension includes: the cognitive error frequency score obtained based on the Cognitive Failure Questionnaire (CFQ); (4) Sleep health dimension characteristic data include: objective physiological sleep indicators: the proportion of deep sleep, REM sleep, light sleep, number of nighttime awakenings, total awake time, and time in bed and total sleep time collected using wearable devices; subjective assessment indicators: the scores of seven components based on the Pittsburgh Sleep Quality Index (PSQI) scale, including subjective sleep quality, time to fall asleep, sleep duration, sleep efficiency, sleep disorders, use of hypnotics, and daytime dysfunction; circadian rhythm indicators: the degree of circadian rhythm disorder calculated based on the absolute difference between the midpoint of sleep on weekdays and the midpoint of sleep on rest days.

[0148] 2. Individual behavioral and environmental exposure characteristics were obtained, including: natural environmental exposure parameters (calculated based on the altitude of the place of residence before entering the plateau) altitude difference and time spent in the plateau; daily activity and work-rest load parameters: daily work / study time, monthly frequency of irregular work and rest, etc. Step S2: Preprocessing of multidimensional heterogeneous data Because physiological, psychological, cognitive, and sleep indicators have different dimensions, and because there are positive indicators (higher values ​​are better, such as blood oxygen) and negative indicators (lower values ​​are better, such as heart rate), range standardization is required. Positive indicator calculation formula: Formula for calculating negative indicators: The range of values ​​is: ; Indicates the first The first subject was on the The original observed values ​​of each evaluation index ( ; ), This represents its dimensionless value after standardization. and These represent the sample population at the 1st... The maximum and minimum values ​​of each evaluation indicator.

[0149] Step S3: Pre-splitting of exposed features based on K-means++.

[0150] Extracting environmental exposure feature vectors from subjects To overcome the shortcomings of random initialization in traditional algorithms, the K-means++ algorithm is used to maximize the distance between initial cluster centers. Then, the objective function is minimized iteratively. The study population was adaptively divided into three groups: acute exposure group, chronic cumulative high-load group, and conventional steady-state adaptation group. This step aimed to eliminate baseline errors caused by differences in exposure backgrounds, ensuring that subsequent evaluations were conducted under homogeneous exposure conditions. Furthermore, the adaptive and stable division of subjects into different exposure background groups ensured highly targeted and reproducible weighting in subsequent assessments. : Represents the objective function, which is the sum of the squared distances from each sample point to its class center. This is achieved through iteration... minimize; : indicates the first division A subgroup of exposed features; : No. Environmental and lifestyle exposure feature vectors of each subject; : indicates the first The cluster center (centroid) of a subgroup.

[0151] Step S4: First-level CRITIC micro-weighting: Within each subgroup, micro-weights are calculated using the volatility and conflict of the indicators. : Calculate the contrast intensity (i.e., standard deviation): ; Based on the Pearson correlation coefficient among indicators Conflict Computation : ; The combination of contrast intensity and conflict generates micro-weights: .

[0152] This step aims to capture the asynchronous changes between physiological indicators and sleep / cognitive indicators, and to decipher the "masking effect" of physiological compensation on latent damage. : No. The standard deviation of an indicator reflects the intensity of its fluctuation and its discriminative power within a subgroup. : No. within subgroup The average of each indicator. :index With indicators The Pearson correlation coefficient between them is used to characterize information redundancy. : No. Quantitative values ​​of conflict for each indicator; The larger the value, the better the indicator can break through the masking effect of physiological compensation; It is the first The micro-level "weight values" that are ultimately assigned to each indicator.

[0153] Step S5: Second-level EWM macro-weighting The independent composite scores of four dimensions (physiological, psychological, cognitive, and sleep) are used as new state-space variables, and the information entropy is calculated using the entropy weight method. This allows for the calculation of the macro-spatial weights of the four macro-dimensions. .

[0154] 1. Calculate the subjects' overall scores across all dimensions. : Based on the micro-indicator weights obtained in step S3 Compared with the standardized values ​​after step S1 Calculate the first The first subject was on the Overall score across major dimensions : , in, For the first Dimensions ( The score includes the number of indicators. This score reflects an individual's performance level in a single dimension, such as physiological function, psychological and emotional state, cognitive efficacy, or sleep health.

[0155] 2. Calculate the weighting of dimensional scores. : To calculate the entropy value, the individual score needs to be normalized within the sample pool of that subgroup to obtain the first value. The first subject was on the Weight of scores in each dimension : , Where m is the total number of samples in the subgroup.

[0156] 3. Calculate the information entropy of each dimension. : The information entropy formula is used to measure the dispersion of data distribution across four dimensions: , in, It is the first Information entropy in a large dimension The smaller the value, the more information that dimension provides, and the higher its weight in the overall health assessment.

[0157] 4. Calculate the macroscopic spatial weights , in, These correspond to the macroscopic spatial weights of the four dimensions: physiological, psychological, cognitive, and sleep. This step aims to prevent highly fluctuating physiological data from masking the evaluative effectiveness of the psychological, cognitive, and sleep dimensions, ensuring the balance of the four-dimensional space, starting from the purity of information.

[0158] Step S6: TOPSIS Spatial Ranging and Adaptive Grading 1. Construct a weighted normalized matrix and extract the ideal solution: The system is based on the individual's comprehensive score and macro weights across all dimensions. Automatically extract the positive ideal solution (population optimal state vector) of the subject group under the current exposure background. ) and negative ideal solution (population range state vector) ).

[0159] 2. Calculate the health relevance score. : Calculate the individual state vector and the positive ideal solution ( and negative ideal solution ( Euclidean distance of ) , in , The larger the value, the closer the individual's overall functional state is to the ideal steady state.

[0160] 3. Adaptive grading based on K-means++: Calculate the set of individual health proximity scores Then, set the number of clusters. (These correspond to pathological risk, functional decline in sub-health, and homeostasis of health, respectively).

[0161] Cluster center initialization: from the score set A sample point is randomly selected as the first initial cluster center. For the set... Any remaining samples Calculate the Euclidean distance from the current nearest existing cluster center. Calculate the probability that each sample will be selected as the next new cluster center. The probability With distance It is directly proportional to the square of, and the formula for its calculation is: , Based on the stated probability The next new cluster center is selected using a roulette wheel method. This process is repeated until all cluster centers have been selected. Initial cluster centers. This step effectively avoids the pitfall of traditional random initialization, which easily gets trapped in local optima. Iterative optimization of the objective function: (1) Iterative clustering and centroid update: Calculate the set Each sample point in the above The Euclidean distance between cluster centers is calculated, and each sample point is assigned to the cluster with the smallest distance. After the division, the centroid of each cluster is updated. The updated formula is as follows ,in For the first Sample size within each cluster The score is given to the samples within the cluster.

[0162] By iteratively dividing the samples and updating the centroids, the sum of squared errors within the cluster is minimized. Until the algorithm converges. The sum of squared errors within the cluster. The objective function is: .

[0163] (2) Dynamic hierarchical boundary threshold extraction: After the algorithm converges, the final centroids corresponding to the three clusters are extracted and sorted by numerical value, and are respectively denoted as the centroids of the "pathological risk" cluster. "Sub-health functional decline" cluster center and the "healthy homeostasis" cluster centroid By calculating the arithmetic mean of adjacent centroids, the hierarchical boundary threshold for generating the current data context is extracted. Calculate the pathological risk boundary threshold : , Calculate the healthy steady-state boundary threshold : .

[0164] The final output is a dynamic evaluation standard sequence. It is used to adaptively grade and warn about the physical and mental health status of subjects.

[0165] At the application level, this step is divided into two stages: "benchmark extraction" and "individual mapping," and has a dynamic update mechanism. Benchmark extraction phase: Generating hierarchical boundary thresholds using initial subject population data. and This serves as an evaluation benchmark within this environmental context; Individual mapping stage: For newly added subjects, their proximity score is directly assigned. With the already generated , Comparisons can be made without re-triggering clustering calculations; Step S7: Sample-driven iterative update mechanism.

[0166] The cumulative sample size in the multidimensional dataset is monitored in real time. When the current cumulative sample size reaches an integer multiple of the initial baseline sample size (trigger mechanism), the calculation and clustering iteration process in steps S2 to S6 is re-executed to update the evaluation criteria for the plateau population. and This will allow us to obtain an updated health status classification model for people living in high-altitude areas, enabling adaptive evolution of the standard.

[0167] Differentiated proactive health intervention strategies based on grading results: The system uses a model-based hierarchical assessment and outputs differentiated intervention instructions based on the "weakest link effect" of multidimensional scores. 1. Status determination triggering mechanism: when When the condition is determined to be "healthy and stable", the system will only perform routine recording and will not trigger intervention.

[0168] when At that time, it was determined to be "sub-health functional decline".

[0169] when When the condition is identified as "pathological risk," the system triggers the highest level of medical warning.

[0170] 2. Identifying the root causes of weaknesses and implementing multi-dimensional coordinated interventions: For subjects who triggered the intervention, the system compared their weighted independent scores across four dimensions: physiological, psychological, cognitive, and sleep. Identify the "weakness dimension" with the lowest score (i.e., ), and map the corresponding execution strategy: If the disadvantage is physiological function: the system outputs "active medical repair" instructions to the terminal, such as increasing the frequency of oxygen inhalation, forcing rest, and forcibly reducing the physical activity load.

[0171] If the disadvantage dimension is psychological emotion or cognitive performance: the system outputs "psychological adjustment" instructions, such as arranging mindfulness relaxation training, suspending high cognitive load activities, etc.

[0172] If the disadvantage dimension is sleep health: the system triggers sleep intervention commands, including outputting "physical environment control strategies" to the smart hardware terminal (such as increasing the ventilation rate of the sleep environment, reducing the carbon dioxide concentration of the sleep environment, increasing the light blocking of the sleep environment, controlling the noise of the sleep environment, etc.), or outputting health suggestions to the management terminal (such as suggesting optimization of the sleep cycle, etc.). Example 2: A Health Status Grading Assessment System for Plateau Population Based on a Four-Dimensional Functional State Space Referring to Example 1, the present invention also proposes a health status grading assessment system for plateau populations based on a four-dimensional functional state space, comprising the following functional modules: (M1) Input module: The input module is configured to input data, which includes a multidimensional dataset of the test subject (i.e., an individual from the plateau population). This module is used to execute method step S1. The module further includes: The physical signs and sleep data acquisition unit collects objective physical signs data such as blood oxygen saturation and multidimensional heart rate variability (time domain, frequency domain, and nonlinear indicators) of the target subjects through wearable devices (such as smart bracelets and dynamic ECG patches) and portable medical instruments; it also collects objective sleep health data such as the proportion of deep sleep, REM sleep, light sleep, number of nighttime awakenings, total awake time, time in bed, and total sleep duration; and calculates the midpoint of sleep on weekdays and rest days based on the attendance system, outputting social jet lag (SJL) to quantify the degree of rhythm disorder.

[0173] Environmental and individual behavioral exposure acquisition: The altitude difference, time spent in high-altitude areas, and daily activity and rest load parameters (daily work / study time, frequency of irregular work and rest, etc.) are obtained through the management system interface to generate environmental and individual behavioral exposure feature vectors.

[0174] Questionnaire and rhythm input unit: The cognitive failure questionnaire (CFQ), psychological scale scores (GAD-7 and PHQ-9 scores), and PSQI sleep quality scale scores of the subjects are obtained through smart terminals (such as mobile phone applets or tablets).

[0175] (M2) Analysis Module: The analysis module analyzes the input data and obtains the analysis results. This module is the core engine of the system and specifically includes the following units: (m2.1) Grading Unit: The grading unit is configured as a health status grading model for a plateau population. This model is constructed using the method described in Example 1 and grades the subjects based on a multidimensional dataset. This unit integrates the computational logic of method steps (S2) to (S6): Preprocessing and pre-processing engine: Perform range standardization in step S2 and then perform step (S3) to divide the test objects into corresponding homogeneous exposed feature subgroups using the K-means++ algorithm.

[0176] The dual-layer objective computing power weighting engine executes steps S4 and S5, using the CRITIC algorithm to independently calculate the micro-weights of physiological, psychological, cognitive, and sleep indicators within a subgroup (breaking the physiological compensation masking effect), and using the entropy weight method (EWM) to calculate the macro-spatial weights of the four dimensions. Spatial ranging and hierarchical mapping engine: Executes step (S6) of TOPSIS spatial ranging to calculate the comprehensive health proximity of the test object. This unit is equipped with a threshold memory to solidify the pathological risk boundary threshold T1 and the health homeostasis boundary threshold T2 generated by clustering. For the real-time input score of the test object, it is directly compared with the stored thresholds T1 and T2 to quickly obtain the hierarchical result (health homeostasis, sub-health functional decline, or pathological risk).

[0177] (m2.2) Intervention Unit: The intervention unit is configured to execute follow-up strategies based on the grading results. If the subject is in the "pathological risk" or "sub-health functional decline" category, the unit traces back to the lowest-scoring weakness dimension to determine and generate the corresponding intervention strategy: If the dimension with the lowest overall score is physiological function, then an "active medical repair" strategy is generated (such as increasing the frequency of oxygen inhalation, mandatory rest, or mandatory reduction of physical activity load).

[0178] If the dimension with the lowest overall score is psychological and / or cognitive efficacy, then a "psychological adjustment" strategy is generated (such as arranging mindfulness relaxation training, postponing high cognitive load activities, etc.).

[0179] If the dimension with the lowest overall score is sleep health, then a "sleep intervention" strategy is generated (including linking smart hardware to improve the ventilation rate of the sleep environment, reduce carbon dioxide concentration, increase light blocking, control noise, or generate nighttime oxygen supplementation instructions).

[0180] (m2.3) Update Unit: The update unit is configured to monitor the cumulative sample size of the multidimensional dataset input by the (M1) input module. This unit is used to execute step (S7), which automatically calls back and executes the calculation and clustering process of steps (S2)-(S6) when the current cumulative sample size meets the preset trigger mechanism (reaching an integer multiple of the initial baseline sample size) to update the boundary thresholds T1 and T2 in the memory, so that the (m2.1) grading unit can always be configured to update the plateau population health status grading model to adapt to the latest population characteristics.

[0181] (M3) Output Module: The output module is configured to output the results of the analysis module. This module distributes the graded early warning conclusions generated in (m2.1) and the differentiated proactive health intervention strategies generated in (m2.2) to different execution terminals to achieve closed-loop management. For example: outputting rating reports and medical repair suggestions to the subject's personal terminal or local medical station; issuing sleep microenvironment physical regulation instructions to the environmental control system; or outputting early warning reports to the management system to optimize the subject's work / study time and circadian rhythm. For example: Medical intervention: Push medical intervention suggestions such as increasing oxygen frequency, mandatory rest, or mindfulness relaxation to the subject's personal terminal or local medical station.

[0182] Psychological and cognitive regulation: Push warning suggestions to the management system and personal terminals to arrange mindfulness relaxation training and forcibly suspend complex and precise operation tasks to prevent human error.

[0183] Sleep health intervention: Sending instructions to the environmental control system or individuals, including linking smart hardware to improve the sleep microenvironment (such as light blocking and noise reduction control), issuing warnings of social jet lag to carry out sleep hygiene education, or suggesting low-flow oxygen supplementation at night.

[0184] Organizational management feedback: Output early warning reports to the management system, suggesting optimization of the frequency of unconventional work and rest schedules or work / study durations of specific subjects to adjust their social jet lag.

[0185] Example 3: Specific application verification and effect analysis of the evaluation method of the present invention To further verify the effectiveness of this invention and the specific execution logic of each algorithm module, this embodiment conducted a model verification and numerical calculation test. It should be specifically noted that the sample data used in this embodiment is only used to intuitively verify the coherence and feasibility of the adaptive pre-diversion and objective combination weighting algorithm logic of this invention through specific numerical calculations. The evaluation method described in this invention highly relies on purely data-driven mathematical measures (such as spatial proximity calculation and dynamic clustering threshold generation). Its underlying computational architecture achieves a high degree of decoupling from the prior norm database of a specific population, no longer relying on fixed static thresholds. Therefore, in practical applications, this method is not limited to the specific altitude or sample distribution shown in this embodiment, but can dynamically generalize and automatically calibrate the evaluation boundary based on the actual characteristics of the input data, possessing broad applicability and robustness under different high-altitude exposure scenarios.

[0186] Based on a typical application scenario in a high-altitude living environment, observational sample data from 30 subjects in the same batch were extracted as the evaluation benchmark group for algorithm derivation. To clearly demonstrate the derivation logic of this invention, this embodiment selects two core indicators each from the dimensions of physiological function, psychological emotion, and sleep health, and one core indicator from the cognitive efficacy dimension as a representative. The typical data calculation process of three of the most representative core subjects (A, B, and C) is explained in detail below: 1. S1: Provides a multidimensional dataset of high-altitude populations. Typical exposure and functional characteristics data were obtained from three core subjects and the remaining samples from the same batch (a total of 30 people). The selected representative indicators and the raw data of the subjects are shown in Table 1.

[0187] Table 1. Raw data of three core subjects Among them, A: climbed 3500m in altitude, stayed for 2 days, worked an average of 10 hours per day, and had no irregular work schedules per month; B: climbed 1500m in altitude, stayed for 200 days, worked an average of 12 hours per day, and had two irregular work schedules per month; C: climbed 1500m in altitude, stayed for 500 days, worked an average of 8 hours per day, and had no irregular work schedules per month.

[0188] 2. S2: Preprocessing of Cube Datasets The system iterates through all samples in the current batch and automatically extracts the global maxima of various functional indicators. ) and minimum value ( (This serves as the evaluation benchmark boundary.) Physiological extremes: Blood oxygenation limit Heart rate limits .

[0189] Psychological extreme value: GAD-7 limit PHQ-9 boundary .

[0190] Cognitive extremes: CFQ limits .

[0191] Sleep extremes: PSQI limits Nighttime wakefulness limits The system performs range standardization mapping according to the formula in Example 1.

[0192] For positive indicators, a formula is used. Taking blood oxygen level B as an example, let's substitute the values ​​into the calculation: .

[0193] For negative indicators, the formula is used. Taking B's CFQ cognitive score and heart rate as an example, let's substitute them into the calculation: ; .

[0194] Similarly, by substituting into the formula, we finally obtained the standardized dimensionless score matrix of the three subjects, A, B, and C (retaining two decimal places, with values ​​ranging from 0 to 1, and higher scores indicating healthier function), as shown in Table 2.

[0195] Table 2. Normalized data of three core subjects Note: In actual system operation, each dimension includes, but is not limited to, the aforementioned complex indicators. This is a brief description of the core logic of the embodiment.

[0196] 3. S3: Pre-processing of environmental exposure features based on clustering algorithm Extract environmental exposure and daily routine data from the subjects and normalize them into exposure feature vectors. , respectively corresponding .

[0197] Based on the consensus in high-altitude medicine regarding the acclimatization stages of populations (i.e., acute exposure period, chronic attrition period, and routine acclimatization period), the number of clusters is pre-defined. The system uses the K-means++ algorithm for training and ranging. (1) Distance probability initialization and centroid iterative calculation: Three centroids are initialized using a distance probability weighting method to avoid local optima, and samples are continuously assigned to the nearest cluster using standard Euclidean distance. In the update phase of each iteration (until the objective function converges), the system redetermines the centroid coordinates by calculating the arithmetic mean of the feature vectors of all samples within the cluster, based on the intra-cluster partitioning results.

[0198] Substitute into the centroid update formula .

[0199] Taking the final convergence calculation of the chronic cumulative high burden group (GB group, including 10 samples including subject B) as an example: The system extracts the standardized exposure feature vectors of the 10 samples in this group and performs column matrix summation: Similarly, using the eigenvector summation and averaging method described above, the final convergence centroids of the other two groups can be accurately calculated: (Acute exposure group center of gravity): (Conventional steady-state adaptive population center of mass): (2) Individual ranging and triage execution: Taking the standardized exposure feature vectors of three core subjects as an example: Subject A (elevation 3500m, 2 days in high-altitude environment, 10 hours of high-intensity work per day, no irregular schedule): Subject B (elevation 1500m, 200 days in high-altitude area, 12 hours of work per day, irregular work and rest twice a month): Subject C (elevation 1500m, 500 days in high-altitude area, 8 hours of work per day, no irregular work and rest schedule): Calculate the squared standard Euclidean distance from the subject to the centroid of each subgroup ( Taking subject B as an example, calculate the squared distances from her to the three centroids: because The squared distance value ( Based on the principle of minimum distance, Subject B was precisely classified into the "Chronic Cumulative High Load Group (GB Group)". Similarly, after distance measurement by the algorithm, Subject A was classified into the "Acute Exposure Group (GA Group)" and Subject C into the "Routine Steady-State Adaptation Group (GC Group)". Subsequent steps S3 to S5 were performed independently within each subgroup. This process completely eliminated the error of manual subjective prior classification.

[0200] 4. S4: Weighting within the micro-dimension based on objective weighting method (taking the GB group to which subject B belongs as an example) After clustering and triage, the system extracted the independent dimensional data within the GB group (a total of 10 subjects) and used the CRITIC algorithm to calculate the contrast intensity (unbiased standard deviation) of each indicator. ) and conflict (derived from Pearson correlation coefficient) ), and then extract micro-weights.

[0201] (1) Basic data extraction and standardization: Taking the physiological function dimensions of the GB group (with blood oxygen and heart rate as two indicators) as an example, the original data and standardized scores of 10 subjects recorded by the system (combined with the global extreme values ​​of S1: blood oxygen limit [72,99], heart rate limit [58,125]) are shown in Table 3.

[0202] Table 3. Raw data and standardized scores of 10 subjects The mean within the group was calculated as: mean blood oxygen score. mean heart rate score .

[0203] (2) Calculate the contrast intensity (standard deviation) ): Substitute into the standard deviation formula , and then proceed with the derivation: Blood oxygen variance , ; Heart rate variance , .

[0204] (3) Calculate the conflict (correlation coefficient) and ): The internal redundancy of blood oxygen and heart rate sequences was calculated using the Pearson correlation coefficient formula. Substituting the matrix data, the calculation yields: Covariance numerator: ∑≈0.1470; denominator product term: Correlation coefficient .

[0205] Substituting into the conflict formula : Blood oxygen conflict ; Heart rate conflict .

[0206] (4) Extracting information carrying capacity ( ) and final micro weights ( ): Substitute into the formula Assess the overall amount of information: Blood oxygen information: ; Heart rate information content: .

[0207] Substitute into the weight normalization formula : Blood oxygen microweight ; Heart rate microweight .

[0208] Similarly, the above-mentioned internal indicators for the dimensions of psychological and emotional health and sleep health were applied. and Matrix operations. The final micro-weights for the four dimensions within the GB group are fully allocated as follows: Physiological (blood oxygen) Heart rate ); Psychology (GAD-7) PHQ-9 Sleep (PSQI) Number of times of waking up Cognition is a single indicator and is directly counted as... .

[0209] (5) Intra-dimensional score integration: Substitute into the formula Calculate the four-dimensional independent composite score of subject B: Physiological score ; Psychological score ; Cognitive score ; Sleep score .

[0210] At this point, through rigorous calculation of underlying indicators, the current state of subject B was precisely quantified: it exhibited a very strong "physiological compensation masking effect" (physiological score as high as...). However, due to long-term high-intensity exposure, their cognition and sleep have shown severe, hidden impairment, with scores of only [insert scores here]. and ).

[0211] 5. S5: Calculation of Comprehensive Score and Determination of Macro-Spatial Weights (1) Standardization of local range within groups and matrix construction: The four-dimensional comprehensive score matrix calculated in S3 for 10 subjects in the GB group was extracted, and the intra-group extreme values ​​of each dimension were extracted (e.g., the intra-group range of the physiological dimension was 0.967 - 0.420 = 0.547). A positive standardization formula was then used. To avoid zero-value errors in subsequent logarithmic calculations, the dimensionless score matrix of the 10 subjects in the group was calculated, as shown in Table 4.

[0212] Table 4. Raw data and standardized scores of 10 subjects Note: For ease of formatting, the table above is consistently rounded to three decimal places. GB_Sample 1 marked with an asterisk in the table represents the range baseline sample for each dimension; its underlying actual output value is... (Right now This non-zero minimum value ensures that the system can be successfully invoked in subsequent steps. It can perform operations without causing abnormal overflows.

[0213] (2) Calculate the characteristic proportion ( ): By summing the feature scores of each column in the above matrix vertically, the sum of the within-group scores for each dimension is obtained as follows: Total physiological score ; Total psychological score ; Total cognitive score ; Total sleep score .

[0214] Substitute into the characteristic proportion formula Extract the information weight of subject B in each dimension: Physiological specific gravity Psychological weight ; Cognitive proportion Sleep proportion .

[0215] (3) Calculate information entropy ( ) and the coefficient of difference ( ): Traverse the entire group's weight matrix Substitute into the information entropy formula (Among them, sample size) constant coefficients From the perspective of physiological function ( Taking the calculation of information entropy as an example, the physiological proportions of 10 subjects in the group are extracted. Calculate them separately Value. Based on subject B ( For example: .

[0216] Similarly, the same logarithmic product operation was performed on the remaining 9 subjects in the group, and the results were summed to obtain the logarithmic sum of the physiological dimensions for the entire group: .

[0217] Substituting the coefficients, we obtain the final physiological information entropy: .

[0218] Similarly, the information entropy of the other three dimensions can be derived: Psychological entropy Cognitive entropy Sleep entropy .

[0219] Then, substitute the formula for the coefficient of difference. The four-dimensional dispersion index is obtained as follows: Physiological difference coefficient ; Psychological difference coefficient ; Cognitive Difference Coefficient ; Sleep difference coefficient .

[0220] (4) Generate the final macroscopic spatial weights ( ): First, calculate the sum of the difference coefficients across the four dimensions, using this as the denominator for weight normalization: .

[0221] Substitute into the normalization formula The system ultimately outputs the macroscopic spatial weighting vector for the GB group: Physiological function weight Psychological and emotional weighting ; Cognitive efficacy weight Sleep health weight .

[0222] This step, through the rigorous and traceable calculations described above, objectively demonstrates that the target population exhibits the greatest data dispersion (variance coefficient) in the "cognitive efficacy" dimension. (Ranked first), containing the most effective information on individual differences. Based on this, the algorithm adaptively compressed the weights of the physiological dimensions that showed convergent performance (reducing them to...). The weight of the cognitive dimension representing differences in latent damage was increased (to a higher level). This completely eliminates the evaluation masking effect caused by simple physiological compensation from the mathematical foundation.

[0223] 6. S6: TOPSIS Spatial Ranging and Subgroup-Specific Dynamic Threshold Extraction Combining the extracted macro-weights, the proximity of an individual to the optimal / inferior state of their subgroup is calculated, and the state boundary is dynamically generated. Taking subject B (GB group) as an example, the derivation process is as follows: (1) Constructing the weighted normalization matrix and extracting the ideal solution: The four-dimensional comprehensive score matrix in S3 ( ) and the macroscopic weights extracted by S4 ( Multiply each term by the other to construct a weighted feature matrix. The weighted score of Subject B is: Physiological weighted score ; Psychological weighted score ; Cognitive weighted score ; Sleep weighted score .

[0224] Traverse the GB weighted matrix and extract the positive ideal solution representing the optimal state. ) and the negative ideal solution representing the range state ( ): Positive ideal solution: ; Negative ideal solution: .

[0225] (2) Euclidean distance measurement and proximity ( )calculate: Substituting into the Euclidean distance formula and Calculate the distance index of subject B: Distance to the ideal solution: Distance to negative ideal solution Substitute into the proximity formula The overall health relevance of subject B was determined as follows: .

[0226] Similarly, the closeness scores of the remaining 9 subjects in the GB group were calculated, forming the closeness set of this subgroup. Executing the above logic within both the GA and GC groups yields the following result: A ; C .

[0227] (3) Dedicated dynamic threshold generation: To achieve adaptive grading without prior knowledge, the closeness scores of 10 subjects in the GB group were analyzed. ) Sets of independent K-means++ clustering algorithms (assuming) ).

[0228] During the iterative convergence phase, the system accurately divides these 10 one-dimensional coordinate points into three state clusters, and extracts the final centroid μ by averaging the proximity of samples within the cluster. Pathological risk cluster (1 sample): centroid ; Healthy homeostatic cluster (1 sample): centroid .

[0229] Sub-health functional decline cluster (8 samples, including subject B): .

[0230] Using the formula of the mean of adjacent centroids The system generates adaptive state boundaries specific to the current camp background for the GB group in real time: Pathological risk boundary Healthy homeostasis boundary .

[0231] Similarly, the adaptive boundary of the GA group is calculated as follows: ; GC group adaptive boundary is .

[0232] 7. S7: Sample-driven iterative update mechanism for evaluation criteria To ensure the accuracy of the evaluation criteria in long-term applications, the system incorporates a sample-driven trigger update mechanism.

[0233] Update trigger determination: The system counts in real time the total number of subjects who have completed the evaluation within each subgroup (GA, GB, GC). Set the update step size coefficient to... (As in this embodiment, let) When the cumulative sample size of the current subgroup... Reaching the initial baseline sample size When the number of people increases to an integer multiple (e.g., from 30 to 60, 120, etc.), the system automatically triggers the "Evaluation Criteria Recalibration" command.

[0234] Calculation logic iteration: 1. The system retrieves the full data within this subgroup. Set of health relevance scores of 100 subjects 2. Rerun the K-means++ clustering engine in step (3) of S5 to re-identify cluster centroids on a larger sample base. 3. Using formulas and Calculate and update the new hierarchical boundary threshold for this scenario.

[0235] Technical effect: This "learn as you go" model ensures It can continuously adapt and fine-tune as high-altitude activity data accumulates. For example, as the overall adaptation level of people moving to a certain area improves, the system will automatically identify and slightly shift the threshold, so that the evaluation criteria always align with the current real-world health benchmarks of the population, avoiding the problem of traditional fixed standards becoming ineffective at different times.

[0236] 8. Differentiated proactive health intervention strategies based on triage results Overall relevance of individuals The logic is determined by setting the specific dynamic threshold of its subgroup and then outputting the optimization closed-loop intervention instruction: For subject A (GA group): Decision logic: .

[0237] The system accurately determines the state as "pathological risk" and immediately triggers the highest level of physical intervention, outputting a warning command to the terminal to "forcefully stop high-intensity activities and immediately intervene with medical care".

[0238] For Subject B (GB group): Decision logic: .

[0239] In traditional single physiological assessments, HBV is easily misjudged as healthy due to its relatively good blood oxygen / heart rate compensation. This system, through observational weighting and distance extrapolation, accurately identifies HBV as being in a state of "sub-healthy functional decline," such as... Figure 3 As shown in the multidimensional health assessment radar chart, although Subject B's physiological functions were close to the health baseline due to compensatory mechanisms, there was a significant collapse in his cognitive efficacy dimension. Therefore, targeted intervention strategies were sent to his personal smart terminal or related early warning system: Subject B was advised to postpone high-cognitive-load activities (such as driving, precision construction work, or complex decision-making and planning) to prevent safety risks caused by cognitive impairment. Simultaneously, it was suggested to coordinate with environmental control devices or personal health assistants to adjust daily routines, reduce the frequency of unconventional activities, and push proactive adjustment programs such as mindfulness meditation and high-altitude rhythmic breathing training to effectively alleviate nervous fatigue and accelerate the repair of cognitive and physical functions.

[0240] For subject C (GC group): Decision logic: .

[0241] The system determines that the state is "healthy and stable" and records the current observation data as a normalized health baseline, recommending that the current work and rest patterns be maintained.

[0242] This embodiment demonstrates that even when facing complex groups like Subject B, whose physiological indicators are in a compensatory state but who are generally in a state of sub-health depletion, this invention can completely break through the masking effect through a purely data-driven CRITIC-EWM two-layer combined weighting, and accurately locate the highly concealed "sub-health state" within... Within the dynamic mathematical range, the system completely abandons the constant settings that rely on expert experience. From the underlying mathematical architecture, it has the ability to generalize and be implemented in different altitudes and physical conditions, effectively solving the core pain point of missed assessments and warnings for sub-healthy groups in high-altitude health evaluations.

[0243] Example 4: Evaluation of the model's effectiveness and robustness in large-scale simulations To verify the effectiveness and robustness of the plateau population health status classification assessment method described in this invention under large-scale data input scenarios, this embodiment uses a scheme of "baseline feature extraction combined with large-scale Monte Carlo extension" to conduct systematic simulation tests.

[0244] 1. Prior extraction of features based on the feature distribution of core samples Observational sample data were extracted from 30 subjects in the same batch. Based on the altitude stay duration or physiological exposure characteristics of the target population, they were clustered into several exposure subgroups with independent data distribution attributes (such as the first exposure subgroup, the second exposure subgroup, etc.). The evaluation system of this invention was used to complete the four-dimensional double-layer weighted calculation and proximity calibration of physiological, psychological, cognitive, and sleep aspects. Through statistical analysis, the probability density distribution law of the multidimensional index fluctuation variance and the comprehensive proximity score of the target population was obtained, and these were used as prior statistical parameters for subsequent large-scale simulation tests.

[0245] 2. Large-scale data expansion and testing based on Monte Carlo algorithm To verify the system's generalization ability under massive datasets, this embodiment introduces a Stratified Monte Carlo algorithm for data augmentation based on the extracted prior distribution patterns. According to the distribution weights of each exposed subgroup, a large-scale simulation validation set containing N=500 samples was generated through targeted augmentation.

[0246] Subsequently, to establish an objective comparison benchmark, this embodiment uses a preset multi-dimensional feature threshold boundary to mark the benchmark state of 500 extended samples, forming a verification reference gold standard. Specifically, the benchmark marking rules are set as follows: (1) Pathological risk status marking: When any core physiological parameter (such as blood oxygen, heart rate, etc.) in the multidimensional monitoring data of the sample exceeds the preset safety lower limit / upper limit, or the score of any core psychological scale exceeds the preset clinical warning threshold, an over-boundary alarm is triggered and it is marked as pathological risk. (2) Health homeostasis labeling: When the multidimensional monitoring data of a sample simultaneously meet the full-dimensional compliance conditions (i.e., all physiological parameters are within the preset normal range and all psychological scale scores are within the preset health baseline), it is labeled as health homeostasis. (3) Marking of sub-health functional decline: When the sample data does not trigger the pathological risk over-limit rule and fails to fully meet the conditions for achieving the health homeostasis, it is defined as a transitional state and marked as sub-health functional decline.

[0247] Those skilled in the art should understand that the specific physiological parameter thresholds and scale scores mentioned above are merely specific application examples for calibrating simulation benchmark data in this embodiment. In practical applications, the system can customize the above rules and conditions according to specific altitudes, types of daily activities, and employer management regulations, and this does not limit the scope of protection of the present invention.

[0248] Finally, the calculated continuous scores of 500 cases are input into the exclusive dynamic classification thresholds (i.e., the pathological judgment line and steady-state judgment line corresponding to each subgroup) generated adaptively based on the K-means++ algorithm for each exposure subgroup for automatic judgment, and are aligned and verified with the above-mentioned baseline state marking results.

[0249] 3. Verification Results and Analysis of Beneficial Technical Effects Through comparison and verification, this embodiment outputs the following two core verification charts: (1) Validation using multi-class receiver operating characteristic (ROC) curves: see attached. Figure 4 As shown, with a large-scale expanded data input of N=500, the areas under the curve (AUC) of this evaluation model for predicting the three states of "health homeostasis", "sub-health functional decline", and "pathological risk" are 0.964, 0.884, and 0.824, respectively. This result demonstrates that the computational logic proposed in this invention does not exhibit algorithm failure or overfitting under large data inputs and possesses extremely high risk screening sensitivity.

[0250] (2) Verification of the accuracy of the multi-level confusion matrix: as shown in the appendix. Figure 5 As shown, 500 simulated samples were classified according to the dynamic classification threshold adaptively generated by combining subgroup features in step S5 of this invention. The results show that 409 samples were accurately predicted to fall on the main diagonal of the confusion matrix, and the overall classification accuracy reached 81.8%. This result effectively proves that the exclusive dynamic boundary threshold generated by this invention can not only keenly capture functional decline and pathological states, but also greatly control the false alarm rate and false negative rate of the ineffective system in the extreme environment of high altitude, and has extremely high engineering application feasibility.

[0251] All documents mentioned in this invention are incorporated herein by reference as if each document were individually incorporated by reference. Furthermore, it should be understood that after reading the foregoing teachings of this invention, those skilled in the art can make various alterations or modifications to this invention, and these equivalent forms also fall within the scope defined by the appended claims.

Claims

1. A method for constructing a health status classification model for a high-altitude population, characterized in that, The method includes the following steps: (S1) Provide a multidimensional dataset of a high-altitude population; the multidimensional dataset includes physiological function data, psychological data, cognitive efficacy data, sleep health data, and environmental exposure characteristic data; (S2) Preprocess the multidimensional dataset to obtain a preprocessed multidimensional dataset; (S3) Using a clustering algorithm and by iteratively minimizing the objective function, the plateau population is classified according to the feature vectors of the environmental exposure feature data in the preprocessed multidimensional dataset, thereby dividing the plateau population into multiple subgroups; (S4) In each subgroup, the micro weight of each indicator in the preprocessed multidimensional dataset is determined using the objective weighting method; (S5) Based on the standardized values ​​of each indicator in the preprocessed multidimensional dataset and the micro-weights, determine the comprehensive score of each subject in the plateau population on four dimensions, and then use the objective weighting method to determine the macro-spatial weight of each dimension; wherein, the four dimensions are physiological dimension, psychological dimension, cognitive dimension and sleep dimension; (S6) Determine the state vector of each subject in the plateau population and the extreme state vector of the plateau population based on the comprehensive score and the macroscopic spatial weight; evaluate the distance between the state vector of each subject and the extreme state vector using a distance algorithm, thereby assessing the health closeness of each subject and obtaining a closeness set; classify the closeness set using a clustering algorithm and by iteratively minimizing the objective function; determine the evaluation criteria of the plateau population based on the cluster center of each cluster, thereby obtaining a plateau population health status classification model.

2. The method as described in claim 1, characterized in that, The high-altitude population includes those who have been exposed to the high-altitude environment for a long time and those who have been exposed for a short to medium time. The physiological function data includes: respiratory system data, nervous system data, circulatory system data, and digestive and metabolic data; the nervous system data includes resting heart rate and multidimensional heart rate variability indicators; the multidimensional heart rate variability indicators include: time-domain indicators, frequency-domain indicators, and nonlinear dynamic indicators; The sleep health data includes: objective physiological sleep indicators, subjective assessment indicators, and rhythm indicators; The environmental exposure characteristic data includes natural environmental exposure parameters and daily activity and rest load parameters; the daily activity and rest load parameters include: daily work / study time and monthly frequency of irregular work and rest.

3. The method as described in claim 1, characterized in that, Step (S3) specifically includes the following steps: (s3.1) Using a clustering algorithm, determine the cluster centers of each subgroup based on the environmental exposure feature data in the preprocessed multidimensional dataset; (s3.2) Based on the cluster centers, the plateau population is classified by iteratively minimizing the objective function, thereby dividing the plateau population into multiple subgroups; The subgroups include the acute exposure group, the chronic cumulative high load group, and the conventional steady-state adaptation group.

4. The method as described in claim 1, characterized in that, Step (S6) specifically includes the following steps: (s6.1) Construct the state vector for each subject in the plateau population based on the comprehensive score and the macro-spatial weight; (s6.2) Determine the extreme state vector of the plateau population based on the extreme values ​​of the state vectors of all subjects in each subgroup; (s6.3) Evaluate the distance between the state vector of each subject and the extreme state vector using a distance algorithm; evaluate the overall health closeness of each subject based on the distance, and obtain a closeness set; (s6.4) Using a clustering algorithm, determine the initial cluster centers of each cluster based on the proximity set; (s6.5) Based on the initial cluster centers, the proximity set is divided into multiple clusters, and new cluster centers are determined by iteratively minimizing the objective function; (s6.6) Determine the evaluation criteria for the plateau population based on the new cluster centers of each cluster, thereby obtaining a health status classification model for the plateau population; Step (s6.6) includes: dividing the clusters into "pathological risk" clusters based on the numerical value of the new cluster centers for each cluster. "Sub-health functional decline" cluster center and "healthy homeostasis" cluster center Based on the cluster centers, pathological risk boundary thresholds and health homeostasis boundary thresholds are identified, and a dynamic evaluation standard sequence with the pathological risk boundary thresholds and health homeostasis boundary thresholds as boundaries is formed, thereby obtaining the health status classification model of the plateau population.

5. The method as described in claim 1, characterized in that, The method further includes the following steps: (S7) Monitor the cumulative sample size in the multidimensional dataset in real time; when the cumulative sample size meets the preset triggering mechanism, repeat steps (S2) to (S6) to update the evaluation criteria for the plateau population, thereby obtaining an updated health status classification model for the plateau population.

6. A system for classifying the health status of people living in high-altitude areas, characterized in that, The system includes: (M1) Input module, the input module is configured to input data, the data including a multidimensional dataset of the test object, the multidimensional dataset including physiological function data, psychological data, cognitive efficacy data, sleep health data and environmental exposure characteristic data; (M2) Analysis module, configured to analyze the data and obtain analysis results; the analysis module includes: (m2.1) A grading unit, wherein the grading unit is configured as a plateau population health status grading model, the plateau population health status grading model grading the test object according to the multidimensional dataset to obtain grading results; the plateau population health status grading model is constructed using the method described in claim 1; (M3) Output module, which is configured to output the results of the analysis module.

7. The system as described in claim 6, characterized in that, The analysis module also includes: (m2.2) Intervention unit, the intervention unit is configured to perform the following operations: based on the grading results, if the subject to be tested is at pathological risk or experiencing functional decline in sub-health, then determine the dimension with the lowest comprehensive score among the subjects to be tested, thereby determining the corresponding intervention strategy; Specifically, if the dimension with the lowest overall score is physiological function, an intervention strategy of "proactive medical repair" is provided; if the dimension with the lowest overall score is psychological and / or cognitive efficacy, an intervention strategy of "psychological adjustment" is provided; and if the dimension with the lowest overall score is sleep health, an intervention strategy of "sleep intervention" is provided.

8. An electronic device comprising a processor and a memory, characterized in that, The memory contains multiple executable instructions, and the processor is used to read the instructions and execute steps in a method for grading and assessing the health status of a test object; the method includes the following steps: (1) Provide a multidimensional dataset of a test subject, the multidimensional dataset including physiological function data, psychological data, cognitive efficacy data, sleep health data and environmental exposure characteristic data; (2) Preprocess the multidimensional dataset to obtain a preprocessed multidimensional dataset; (3) Based on the feature vectors of the environmental exposure feature data in the preprocessed multidimensional dataset, the test object is matched to a predetermined exposure feature subgroup; based on the predetermined micro-weights in the exposure feature subgroup to which the test object belongs and the standardized values ​​of each indicator in the preprocessed multidimensional dataset, the comprehensive score of the test object is determined; based on the comprehensive score and the predetermined macro-spatial weights, the state vector of the test object is determined; the distance between the state vector of the test object and the predetermined extreme state vector is evaluated by a distance algorithm, thereby evaluating the comprehensive health proximity of the test object; (4) Input the comprehensive health closeness of the test subject into the plateau population health status classification model and compare it with the predetermined boundary threshold therein, so as to classify the test subject; The boundary thresholds include pathological risk boundary thresholds. and healthy steady-state boundary threshold ; If the overall health relevance of the test subject is... If the overall health closeness of the tested object is [value missing], it indicates that the object is in a healthy homeostasis state; if the overall health closeness of the tested object is [value missing], it indicates that the object is in a healthy homeostasis state. If the test subject is at pathological risk, then the overall health proximity C of the test subject satisfies the following conditions. If so, it indicates that the subject is in a state of sub-health and functional decline. The health status classification model for the plateau population is constructed using the method described in claim 1.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when read and executed by a processor, implement steps in a method for grading and assessing the health status of an object under test; the method includes the following steps: (1) Provide a multidimensional dataset of a test subject, the multidimensional dataset including physiological function data, psychological data, cognitive efficacy data, sleep health data and environmental exposure characteristic data; (2) Preprocess the multidimensional dataset to obtain a preprocessed multidimensional dataset; (3) Based on the feature vectors of the environmental exposure feature data in the preprocessed multidimensional dataset, the test object is matched to a predetermined exposure feature subgroup; based on the predetermined micro-weights in the exposure feature subgroup to which the test object belongs and the standardized values ​​of each indicator in the preprocessed multidimensional dataset, the comprehensive score of the test object is determined; based on the comprehensive score and the predetermined macro-spatial weights, the state vector of the test object is determined; the distance between the state vector of the test object and the predetermined extreme state vector is evaluated by a distance algorithm, thereby evaluating the comprehensive health proximity of the test object; (4) Input the comprehensive health closeness of the test subject into the plateau population health status classification model and compare it with the predetermined boundary threshold therein, so as to classify the test subject; The boundary thresholds include pathological risk boundary thresholds. and healthy steady-state boundary threshold ; If the overall health relevance of the test subject is... If the overall health closeness of the tested object is [value missing], it indicates that the object is in a healthy homeostasis state; if the overall health closeness of the tested object is [value missing], it indicates that the object is in a healthy homeostasis state. If the test subject is at pathological risk, then the overall health proximity C of the test subject satisfies the following conditions. If so, it indicates that the subject is in a state of sub-health and functional decline. The health status classification model for the plateau population is constructed using the method described in claim 1.

10. A computer program product comprising computer-executable instructions, characterized in that, The computer-executable instructions, when executed by a processor, implement the steps in a method for grading and assessing the health status of a test object; the method includes the following steps: (1) Provide a multidimensional dataset of a test subject, the multidimensional dataset including physiological function data, psychological data, cognitive efficacy data, sleep health data and environmental exposure characteristic data; (2) Preprocess the multidimensional dataset to obtain a preprocessed multidimensional dataset; (3) Based on the feature vectors of the environmental exposure feature data in the preprocessed multidimensional dataset, the test object is matched to a predetermined exposure feature subgroup; based on the predetermined micro-weights in the exposure feature subgroup to which the test object belongs and the standardized values ​​of each indicator in the preprocessed multidimensional dataset, the comprehensive score of the test object is determined; based on the comprehensive score and the predetermined macro-spatial weights, the state vector of the test object is determined; the distance between the state vector of the test object and the predetermined extreme state vector is evaluated by a distance algorithm, thereby evaluating the comprehensive health proximity of the test object; (4) Input the comprehensive health closeness of the test subject into the plateau population health status classification model and compare it with the predetermined boundary threshold therein, so as to classify the test subject; The boundary thresholds include pathological risk boundary thresholds. and healthy steady-state boundary threshold ; If the overall health relevance of the test subject is... If the overall health closeness of the tested object is [value missing], it indicates that the object is in a healthy homeostasis state; if the overall health closeness of the tested object is [value missing], it indicates that the object is in a healthy homeostasis state. If the test subject is at pathological risk, then the overall health proximity C of the test subject satisfies the following conditions. If so, it indicates that the subject is in a state of sub-health and functional decline. The health status classification model for the plateau population is constructed using the method described in claim 1.

Images