Clinical aging comprehensive assessment method and system and storage medium

Abstract

The application relates to the technical field of data processing, and discloses a clinical aging comprehensive evaluation method, a system and a storage medium. The method comprises the following steps: collecting and processing physiological indexes to obtain physiological index data; performing cognitive function testing and compensation based on the physiological index data to obtain cognitive evaluation data; analyzing behavior characteristics according to the physiological indexes and the cognitive evaluation data to obtain activity characteristic data; performing matrix processing on the three types of data to obtain aging characteristic data; performing scoring and recombination on the aging characteristic data to obtain aging quantitative data; and formulating and optimizing an intervention scheme according to the aging quantitative data. The application improves the accuracy of the clinical aging comprehensive evaluation method.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data processing, and particularly relates to a clinical aging comprehensive evaluation method and system and a storage medium. BACKGROUND

[0002] With the acceleration of global population aging, the research on aging evaluation and intervention methods is increasingly important. Existing aging evaluation methods mainly include three aspects of physiological index monitoring, cognitive function testing and behavior characteristic analysis. In the aspect of physiological index monitoring, basic data such as blood pressure, heart rate and blood glucose are obtained through wearable devices and medical examinations; in the aspect of cognitive function testing, standardized scales are used to evaluate cognitive abilities such as memory, attention and reaction speed; in the aspect of behavior characteristic analysis, social activities and daily life abilities are evaluated through questionnaire survey and activity record. These evaluation methods provide basic data support for aging research.

[0003] However, the existing technology has the following deficiencies: first, the evaluation indexes are independent of each other, lacking systematic integration analysis, resulting in one-sided evaluation results that cannot fully reflect the individual's aging state; second, the influence of human physiological rhythm is not considered in the evaluation process, and the test results are easily disturbed by time factors; third, the data collection and processing method is relatively simple, and cannot effectively handle data noise and outliers, affecting the accuracy of the evaluation results; finally, there is a lack of personalized intervention scheme optimization mechanism based on the evaluation results, making it difficult to provide precise health management suggestions for the elderly. SUMMARY

[0004] The present application provides a clinical aging comprehensive evaluation method, system and storage medium, which is used to establish a multi-dimensional and dynamic clinical aging comprehensive evaluation method, realize the systematic integration of physiological indexes, cognitive functions and behavior characteristics, and provide accurate aging evaluation results and personalized intervention schemes through data mining and machine learning technologies.

[0005] In a first aspect, the application provides a clinical aging comprehensive evaluation method, which comprises: collecting physiological index data by data denoising and normalization processing; performing period test on cognitive function based on the physiological index data, performing compensation processing according to a rhythm curve, and obtaining cognitive evaluation data by data fusion analysis; tracking and recording behavior characteristics according to the physiological index data and the cognitive evaluation data, and obtaining activity characteristic data by social frequency calculation and network structure analysis; performing matrix construction and data decomposition on the physiological index data, the cognitive evaluation data and the activity characteristic data, and obtaining aging characteristic data by threshold calculation and data reconstruction; performing score calculation and feature recombination on the aging characteristic data, and obtaining aging quantitative data by baseline comparison and weight adjustment; decomposing an intervention scheme into targets according to the aging quantitative data, and obtaining an optimized intervention scheme by effect evaluation and parameter adjustment.

[0006] In a second aspect, the application provides a clinical aging comprehensive evaluation system, which comprises:

[0007] A collection module is configured to collect physiological index data by data denoising and normalization processing.

[0008] A compensation module is configured to perform period test on cognitive function based on the physiological index data, perform compensation processing according to a rhythm curve, and obtain cognitive evaluation data by data fusion analysis.

[0009] An analysis module is configured to track and record behavior characteristics according to the physiological index data and the cognitive evaluation data, and obtain activity characteristic data by social frequency calculation and network structure analysis.

[0010] A decomposition module is configured to perform matrix construction and data decomposition on the physiological index data, the cognitive evaluation data and the activity characteristic data, and obtain aging characteristic data by threshold calculation and data reconstruction.

[0011] A recombination module is configured to perform score calculation and feature recombination on the aging characteristic data, and obtain aging quantitative data by baseline comparison and weight adjustment.

[0012] An adjustment module is configured to decompose an intervention scheme into targets according to the aging quantitative data, and obtain an optimized intervention scheme by effect evaluation and parameter adjustment.

[0013] In a third aspect, the application provides a computer readable storage medium, which stores instructions, when the instructions are run on a computer, the computer performs the clinical aging comprehensive evaluation method described above.

[0014] In the technical solutions provided in the present application, the interference factors in the physiological index data are effectively eliminated through data denoising and normalization processing, thereby improving the quality and reliability of the original data. The cognitive function period test based on the rhythm curve overcomes the defects of traditional test methods that ignore the influence of human physiological rhythm, thereby realizing more scientific cognitive ability evaluation. The tracking record of behavior characteristics and network structure analysis break through the limitations of traditional evaluation methods that only focus on single behavior indicators, thereby realizing comprehensive understanding of social activities and behavior patterns. The application of matrix construction and data decomposition technology solves the problem of difficult unified processing of multi-source heterogeneous data, thereby realizing effective integration of physiological, cognitive and behavior data. The score calculation and feature reorganization process overcomes the strong subjectivity of traditional scoring methods, thereby establishing objective and quantitative evaluation standards. The intervention scheme development based on the quantitative data of aging realizes seamless connection from evaluation to intervention, thereby improving the accuracy of clinical comprehensive evaluation of aging. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor based on these drawings.

[0016] Figure 1 An embodiment schematic diagram of the clinical comprehensive evaluation method of aging in the present application;

[0017] Figure 2 An embodiment schematic diagram of the rhythm curve in the present application;

[0018] Figure 3 An embodiment schematic diagram of the clinical comprehensive evaluation system of aging in the present application. DETAILED DESCRIPTION

[0019] The embodiment of the present application provides a clinical aging comprehensive evaluation method, system and storage medium. The terms "first", "second", "third", "fourth" and the like (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not have to be used to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" or "have" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device including a series of steps or units does not have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0020] For ease of understanding, the specific process of the embodiment of the present application is described below. Please refer to Figure 1 One embodiment of the clinical aging comprehensive evaluation method in the embodiment of the present application comprises the following steps.

[0021] Step S101, data acquisition is performed on physiological indexes, physiological index data is obtained through data denoising and normalization processing;

[0022] Step S102, period test is performed on cognitive function based on the physiological index data, compensation processing is performed according to the rhythm curve, and cognitive evaluation data is obtained through data fusion analysis;

[0023] Step S103, behavior characteristics are tracked and recorded according to the physiological index data and the cognitive evaluation data, activity characteristic data is obtained through social frequency calculation and network structure analysis;

[0024] Step S104, matrix construction and data decomposition are performed on the physiological index data, the cognitive evaluation data and the activity characteristic data, and aging characteristic data is obtained through threshold calculation and data reconstruction;

[0025] Step S105, score calculation and feature reorganization are performed on the aging characteristic data, and aging quantitative data is obtained through baseline comparison and weight adjustment;

[0026] Step S106, target decomposition is performed on the intervention scheme according to the aging quantitative data, and the intervention optimization scheme is obtained through effect evaluation and parameter adjustment.

[0027] It can be understood that the execution subject of the present application can be a clinical aging comprehensive evaluation system, and can also be a terminal or a server, and the specific place is not limited. The embodiment of the present application takes the server as the execution subject for example.

[0028] Specifically, physiological indicators are collected by a multi-channel physiological sensor array. The collected indicators include basic physiological data such as blood pressure, heart rate, blood glucose, and cholesterol levels. Data collection adopts two methods: real-time monitoring and periodic detection. Real-time monitoring continuously records dynamic indicators such as heart rate and blood pressure through wearable devices such as smartwatches; periodic detection obtains static indicators such as blood glucose and cholesterol through medical examinations. The collected raw data is denoised by wavelet transform to eliminate random noise and baseline drift generated during measurement, and then normalized by maximum and minimum values to map indicators of different dimensions to a unified interval [0, 1], forming standardized physiological indicator data. Based on the standardized physiological indicator data, individuals are tested for cognitive function in different time periods according to the circadian rhythm. The test content covers multiple dimensions such as memory, attention, and reaction speed, and the test period is divided into three time windows: morning, noon, and evening. During the test, the test results are dynamically compensated by the physiological rhythm curve to eliminate the influence of time factors on cognitive performance. Specifically, the morning 7-9 is the active period of cognitive function, and the test results need to be multiplied by a compensation coefficient of 0.9; the noon 12-14 is the cognitive function trough period, and the test results need to be multiplied by a compensation coefficient of 1.2; the evening 17-19 is the sub-active period of cognitive function, and the test results need to be multiplied by a compensation coefficient of 1.0. Through multi-level data fusion analysis, the compensated test data is integrated into unified cognitive evaluation data.

[0029] On the basis of obtaining physiological indicator data and cognitive evaluation data, the behavior characteristics of individuals are recorded all day long. Through smart terminals, the activity trajectory, social media usage, and exercise data of individuals are recorded. In the social frequency calculation link, the number of daily social activities, single social duration, and the number of social objects are calculated. Network structure analysis calculates the node degree centrality, betweenness centrality, and other topological features by constructing a social relationship network graph to quantitatively describe the breadth and depth of the social network. Through these analyses, activity characteristic data is ultimately obtained. For the three types of data obtained: physiological indicator data, cognitive evaluation data, and activity characteristic data, a three-order tensor matrix is first constructed to uniformly represent the characteristics of different dimensions. The characteristic tensor is decomposed by high-order singular value decomposition to extract the main characteristic components. In the threshold calculation link, a density-based clustering method is used to determine the feature selection threshold to eliminate redundant and noisy features. Finally, the selected features are reconstructed by non-negative matrix factorization to obtain aging characteristic data reflecting the aging state of individuals.

[0030] The quantitative scoring calculation of aging characteristic data is first analyzed by probability density estimation to establish the scoring standard based on age stratification. During the feature reconstruction process, principal component analysis is used to reconstruct the feature space and extract the most representative feature combination. In the baseline comparison section, individual scores are compared with the baseline of the same age group, and the confidence interval of the score is estimated by Bootstrap sampling. The weight adjustment uses the entropy weight method to dynamically allocate weight coefficients according to the information size of the features, and finally obtains the aging quantitative data. According to the aging quantitative data, individualized intervention programs are developed, and the intervention targets are decomposed by analytic hierarchy process to establish an intervention system containing multiple dimensions such as exercise prescription, nutrition conditioning, and cognitive training. In the effect evaluation stage, gray correlation analysis is used to calculate the correlation between intervention measures and effect indicators. In the parameter adjustment process, genetic algorithm is used to optimize intervention parameters such as exercise intensity, training duration, and nutrient supplement dosage, and finally a dynamically optimized intervention program is formed.

[0031] For example: A medical institution conducts a three-month aging assessment and intervention on a 65-year-old old man. In the physiological data collection stage, it is recorded that the old man's average blood pressure is 135 / 85 mmHg, resting heart rate 72 times / min, fasting blood glucose 5.8 mmol / L, and total cholesterol 5.2 mmol / L. Cognitive function tests show that his cognitive scores at different times are: 85 points in the morning (76.5 points after compensation), 65 points at noon (78 points after compensation), and 75 points in the evening (75 points after compensation). Behavior tracking records show that he participates in social activities 4-5 times a week, with an average duration of 2 hours each time, and his fixed social circle includes 8 people. Through data processing and analysis, the old man's aging characteristic score is finally calculated as 3.2 points (full score 5 points), indicating that he is in a mild aging state. Based on this evaluation result, an intervention program is developed, including 3 times of moderate-intensity aerobic exercise per week (30 minutes each time), cognitive game training (1 hour per day), and social activity arrangement (increasing 1 group activity per week). After three months of intervention, his aging characteristic score decreased to 2.8 points, showing good intervention effect.

[0032] In the embodiments of the present application, through data denoising and normalization processing, the interference factors in the physiological indicator data are effectively eliminated, and the quality and reliability of the original data are improved; based on the rhythm curve cognitive function period test, the defects of ignoring the influence of human physiological rhythm in the traditional test method are overcome, and more scientific cognitive ability evaluation is realized; the tracking record and network structure analysis of behavior characteristics break through the limitation of traditional evaluation methods which only focus on single behavior indicators, and realize the comprehensive grasp of social activities and behavior patterns; the application of matrix construction and data decomposition technology solves the problem of difficult unified processing of multi-source heterogeneous data, and realizes the effective integration of physiological, cognitive and behavior data; the process of score calculation and feature reorganization overcomes the subjective shortcomings of traditional scoring methods, and establishes an objective and quantitative evaluation standard; the intervention scheme based on the quantitative data of aging is formulated, realizing the seamless connection from evaluation to intervention, and improving the accuracy of clinical comprehensive evaluation of aging.

[0033] In a specific embodiment, the process of step S101 can specifically include the following steps:

[0034] (1) The multi-channel physiological sensor array classifies and collects heart rate, blood pressure and blood glucose according to different time windows, and obtains an original physiological data set after data pre-screening and time sequence completion;

[0035] (2) According to the frequency spectrum characteristics of the original physiological data set, the high-frequency noise and baseline drift are separated by adaptive wavelet decomposition, and the artifacts are eliminated by combining empirical mode decomposition, to obtain a denoising data set;

[0036] (3) Periodic fluctuation features are extracted from the denoising data set, and the data is stabilized by combining weighted moving average and nonlinear smoothing technology, and the trend items are removed by piecewise regression, to obtain a smoothed data;

[0037] (4) A multi-dimensional feature space is constructed for the smoothed data, and data dimensionality reduction and reconstruction are performed by principal component rotation and maximum information entropy criterion, and standardized feature data is obtained by combining Z-score standardization processing;

[0038] (5) An automatically adjusted dynamic threshold boundary is set for the standardized feature data, and outliers are identified by density clustering and anomaly detection algorithm, and data correction is performed by combining expert rules, to obtain corrected data;

[0039] (6) The corrected data is grouped by hierarchical clustering, and key features are extracted according to the information gain criterion, and physiological indicator data is obtained after feature combination and data alignment.

[0040] Specifically, in the process of physiological indicator data collection and processing, first of all, data collection is carried out by a multi-channel physiological sensor array. The sensor array includes a blood pressure sensor, a heart rate sensor and a blood glucose sensor, and different time windows are adopted for hierarchical collection. Blood pressure data is collected every 30 minutes, heart rate data is collected every 5 minutes, and blood glucose data is collected every 2 hours. In the data pre-screening stage, validity test is adopted to eliminate obvious outliers, such as data with heart rate exceeding 40-120 times / minute. For the data missing in the collection process, linear interpolation is used to fill in the time series, ensuring the continuity of the data, so as to obtain the original physiological data set. After obtaining the original physiological data set, noise processing is needed. First, the spectrum of the data is analyzed to determine the main frequency components of the signal. Adaptive wavelet decomposition technology is used to select appropriate wavelet basis functions for multi-scale decomposition of the data. For heart rate data, db4 wavelet is selected for 5-layer decomposition, and signal components in different frequency bands are extracted. High-frequency noise is mainly concentrated in the first 2 layers of wavelet coefficients, and is processed by soft threshold denoising method. Baseline drift is mainly reflected in the 4th-5th layer wavelet coefficients, and is corrected by median filtering. For the artifacts still existing after processing, empirical mode decomposition technology is used for further optimization. Empirical mode decomposition decomposes the signal into a number of intrinsic mode functions, and by analyzing the frequency characteristics of each intrinsic mode function, the artifact component is identified and removed, and finally the denoised data set is obtained.

[0041] When the denoised data set is processed for stability, the periodic fluctuation characteristics of the data are first extracted. The periodic components of the signal are analyzed by Fourier transform to identify the main physiological rhythm characteristics. Then, a weighted moving average technique is used to preliminarily smooth the data, and the window size is adjusted according to the sampling frequency of different indicators, such as 15-minute window for heart rate data and 1-hour window for blood pressure data. On this basis, nonlinear smoothing technology is introduced, and local regression algorithm is used to further process the data, effectively preserving the local characteristics of the data. The long-term trend item of the data is identified by piecewise regression analysis, and is removed from the original data to obtain the smoothed data. When constructing the feature space for the smoothed data, each physiological indicator is represented as a different dimension. The feature space is transformed by principal component rotation technique to find the optimal feature representation direction. The maximum information entropy criterion is used for feature selection to retain the feature combination with the largest information amount. The selected features are subjected to Z-score standardization processing to convert different dimension indicators to a unified scale space to obtain standardized feature data.

[0042] After obtaining the standardized feature data, it is necessary to perform outlier detection and processing. Set a dynamic threshold boundary, and the threshold is automatically adjusted according to the data distribution characteristics. Adopt DBSCAN density clustering algorithm to perform clustering analysis on the data, and identify outliers in the data. Combine the physiological index normal range rules formulated by experts to correct the identified outliers. For the outliers that cannot be determined, adopt local mean substitution to obtain the corrected data. Finally, perform feature extraction and organization on the corrected data. Adopt hierarchical clustering algorithm to group the data, and construct a hierarchical tree structure according to the similarity between the data. Evaluate the importance of different features through information gain criterion, and select the feature combination with the most value for aging evaluation. Align the selected features in time sequence to obtain the final physiological index data.

[0043] For example: A medical institution collects and processes the physiological data of a 70-year-old person. In 24 hours of continuous monitoring, the heart rate sensor collects data every 5 minutes, a total of 288 data points, and the original heart rate range is between 55-95 times / minute. After removing measurement noise through adaptive wavelet decomposition, the heart rate data range is narrowed to 58-88 times / minute. After weighted moving average processing (15-minute window), a smoother heart rate change curve is obtained. Blood pressure data is collected every 30 minutes, a total of 48 data points, and the original systolic pressure range is between 115-165 mmHg. After noise reduction and smoothing processing, the blood pressure data range is adjusted to 125-155 mmHg. Blood glucose data is collected every 2 hours, a total of 12 data points, and the value range is between 4.8-7.2 mmol / L. Through feature space construction and standardization processing, the three groups of indexes are unified to the standard normal distribution space. Density clustering analysis finds two outlier points (heart rate 92 times / minute, blood pressure 162 mmHg), and after expert rule correction, more reasonable values are obtained (heart rate 85 times / minute, blood pressure 150 mmHg). Through feature extraction and combination, a complete data set reflecting the 24-hour physiological state changes of the old person is formed.

[0044] In a specific embodiment, the process of performing step S102 can specifically include the following steps:

[0045] (1) Associate and pair the physiological index data with the time stamp, extract the physiological fluctuation rule through the circadian rhythm analysis algorithm, and obtain the rhythm reference data;

[0046] (2) Establish a multi-dimensional cognitive test matrix for the rhythm reference data, perform task difficulty grading and time window division, combine the preset rhythm curve and attention fluctuation curve, and obtain test sequence data;

[0047] (3) Set a dynamic response threshold for the test sequence data, perform test result quantification through reaction time statistics and error mode analysis, and obtain original cognitive data;

[0048] (4) Extracting the timing features from the original cognitive data, separating the cognitive fluctuation patterns through wavelet packet decomposition, and evaluating the complexity through entropy calculation to obtain cognitive feature data;

[0049] (5) Multi-scale analysis of cognitive feature data, extracting instantaneous phase information through Hilbert transform, and combining rhythm compensation algorithm for data correction to obtain compensated data;

[0050] (6) The compensated data is processed through multi-level data fusion, combined with information entropy weight distribution for feature integration, and the multi-dimensional features are adaptively combined to obtain cognitive evaluation data.

[0051] Specifically, the physiological index data is associated with the time marker for pairing, and a time series database is established. The 24-hour variation of physiological indicators is analyzed through circadian rhythm analysis algorithm, including the key periods of morning body temperature rise period (06:00-08:00), noon body temperature stable period (12:00-14:00) and night body temperature drop period (20:00-22:00) etc. The rhythm reference data reflects the periodic variation characteristics of human physiological indicators, providing time reference for subsequent cognitive test. After obtaining the rhythm reference data, a multi-dimensional cognitive test matrix is established, including memory test, attention test and reaction speed test. The task difficulty is divided into three levels: primary, intermediate and advanced. The primary task includes simple number memory (3-5 digits), pattern recognition (2-3 targets); the intermediate task includes medium complexity word memory (5-7 words), multi-target tracking (4-6 targets); the advanced task includes complex logical reasoning, multi-task coordination, etc. The time window is divided considering the rhythm curve of human body, as shown in Figure 2 The curve shows that the cognitive ability of human body presents obvious fluctuation in a day: the first peak period of cognitive ability is 8:00-10:00 in the morning, the excitability of cerebral cortex is high, which is suitable for complex cognitive tasks; the trough period of cognitive ability is 12:00-14:00 in the afternoon, when the brain needs rest; the second peak period of cognitive ability is 15:00-17:00 in the afternoon, which is suitable for medium difficulty cognitive tasks. The attention fluctuation curve shows the variation law of human attention level: the attention level gradually rises in the morning, reaching the peak at 9:00-11:00, with a duration of about 2 hours; there is a short-term decrease in attention at 14:00-15:00 in the afternoon; there is a second peak of attention at 18:00-20:00 in the evening. Combined with the characteristics of the two curves, the test tasks are arranged in the most suitable time window to form a scientific test sequence data.

[0052] The dynamic response threshold is set for the test sequence data, and the threshold is dynamically adjusted according to the test difficulty and time window. The reaction time statistics adopt a double recording method, which records both the first reaction time and the total time to complete the task. The error mode analysis includes three types of errors: omission errors (missing target stimuli), false reaction errors (reacting to non-target stimuli), and reaction retardation (exceeding the normal reaction time range). Weighted calculation is performed for each error type to obtain the original cognitive data. From the original cognitive data, the time sequence features are extracted through wavelet packet decomposition, and the cognitive fluctuation pattern is decomposed into components of different frequency bands. Wavelet packet decomposition can accurately capture the subtle changes in cognitive performance and distinguish between short-term fluctuations and long-term trends. By calculating the entropy value of each frequency band, the complexity and stability of cognitive performance are evaluated, and finally the cognitive feature data is obtained.

[0053] The cognitive feature data is subjected to multi-scale analysis, and the instantaneous phase information of the data is extracted using Hilbert transform. Hilbert transform can accurately reflect the dynamic change characteristics of cognitive performance, providing a basis for rhythm compensation. The rhythm compensation algorithm corrects the test results according to the fluctuation of cognitive ability in different time periods to obtain the compensated data. Finally, the compensated data is subjected to multi-level fusion processing. The information entropy weight distribution scientifically allocates weight coefficients according to the information amount of each index. The multi-dimensional features are adaptively combined to form the final cognitive evaluation data.

[0054] For example: A 68-year-old old man is evaluated for cognitive function. At 8:30 in the morning, the average simple reaction time is 285 milliseconds, and the accuracy rate is 92%. Considering that this is a period of rising cognitive ability, the rhythm curve compensation coefficient is 0.9, and the corrected reaction time is 256.5 milliseconds. In the memory test, the immediate recall accuracy rate for a 6-digit sequence is 83%, and the delayed recall (after 30 minutes) accuracy rate drops to 75%. The test at 13:00 shows that the simple reaction time has lengthened to 320 milliseconds, and the accuracy rate has dropped to 85%. At this time, the rhythm compensation coefficient is 1.2, and the corrected reaction time is 384 milliseconds. In the test at 16:00, the reaction time has returned to 290 milliseconds, and the accuracy rate has increased to 90%. Through wavelet packet decomposition, it is found that the main frequency of cognitive fluctuation is concentrated in the range of 0.1-0.3 Hz, indicating that the cognitive function is relatively stable. Information entropy analysis shows that the weight of reaction time index is 0.4, the weight of memory index is 0.35, and the weight of attention index is 0.25. Comprehensive calculation shows that the cognitive function score of the old man is 78 points (full score 100 points), which is at the medium level of the same age group. This score fully considers the fluctuation of cognitive performance throughout the day and eliminates the influence of time factors through rhythm compensation.

[0055] In a specific embodiment, the process of performing step S103 can specifically include the following steps:

[0056] (1) Time alignment of physiological index data and cognitive assessment data, construction of behavior timing benchmark through multi-source data correlation analysis, and obtaining of benchmark feature data;

[0057] (2) Setting multi-dimensional behavior monitoring indicators for benchmark feature data, classifying and counting daily activities through behavior pattern recognition algorithm, and obtaining behavior statistical data;

[0058] (3) Extracting social interaction information from behavior statistical data, quantitatively calculating interaction intensity and frequency through social network topology analysis, and obtaining social feature data;

[0059] (4) Spatiotemporal dimension expansion of social feature data, division of social group structure through spectral clustering algorithm, feature extraction combined with centrality calculation, and obtaining network structure data;

[0060] (5) Social relationship mining of network structure data through graph theory analysis, dynamic tracking combined with social group evolution characteristics, and obtaining social dynamic data;

[0061] (6) Multi-dimensional feature combination of social dynamic data, time series correlation analysis and feature fusion processing, activity feature data obtained combined with behavior pattern evolution law.

[0062] Specifically, the physiological index data and cognitive assessment data are aligned according to the time stamp. The physiological index data includes continuous measurement values such as heart rate and blood pressure, and the cognitive assessment data includes cognitive test results at each time period. Through time series analysis technology, data with different sampling frequencies are unified to the same time scale to construct a unified behavior timing benchmark. The multi-source data correlation analysis uses a sliding time window method, with a window size of 30 minutes and a sliding interval of 15 minutes. The correlation of each indicator within the analysis window is analyzed to obtain the benchmark feature data. Based on the benchmark feature data, a multi-dimensional behavior monitoring indicator system is set up, including basic indicators such as daily activity intensity, activity duration, and activity frequency. The behavior pattern recognition algorithm first classifies the collected behavior data, dividing daily activities into four levels: resting activities (such as watching TV, reading), light activities (such as walking, doing housework), moderate activities (such as fast walking, playing Tai Chi), and heavy activities (such as running, playing ball). The duration and frequency of each type of activity are recorded to generate behavior statistical data.

[0063] Social interaction information is extracted from behavioral statistics, including face-to-face conversations, group activity participation, phone communication, and other forms. Social network topology analysis first constructs a social relationship graph, with nodes representing individuals and edges representing social connections. By recording the duration and number of participants in each social activity, social interaction intensity is calculated. Social frequency is obtained by counting the number of social activities per unit time. Integrating these indicators, social characteristic data is obtained. The spatiotemporal dimensions of social characteristic data are unfolded to analyze the temporal distribution and spatial distribution characteristics of social activities. Spectral clustering algorithm calculates the eigenvalues and eigenvectors of the Laplacian matrix by constructing the adjacency matrix of the social network, and classifies social groups naturally. Centrality calculation includes degree centrality (number of direct social connections), betweenness centrality (degree of social bridge), and closeness centrality (average social distance from others), which together form network structure data.

[0064] Network structure data is further mined through graph theory analysis. First, calculate the basic characteristics of the network, such as average path length and clustering coefficient. Social group evolution characteristics are obtained by tracking changes in the social network at different time points, including new social connections and changes in connection strength, forming social dynamic data. Finally, feature fusion is performed on social dynamic data, combining time series features, network topology features, and behavior pattern features. Time series correlation analysis evaluates the trend of each feature over time, and behavior pattern evolution reflects the long-term change characteristics of social behavior. Through feature weighted fusion, the final activity characteristic data is obtained.

[0065] For example: A 72-year-old old man's behavior characteristics are analyzed for a month. First, align his heart rate data (every 5 minutes) and cognitive test results (3 times a day) in time, and find that the heart rate increases slightly (average increase 8-10 times / minute) during social activities, and the cognitive performance is relatively stable. Behavior pattern recognition shows that daily resting activity accounts for 45% (about 6.5 hours), light activity accounts for 35% (about 5 hours), moderate activity accounts for 15% (about 2 hours), and heavy activity accounts for 5% (about 0.5 hours). Social network analysis shows that the old man has a fixed social circle of 12 people, of which the core interaction circle is 5 people. The average daily social time is 3.2 hours, including morning exercise group activities (1.5 hours), neighborhood conversations (1 hour), and family interactions (0.7 hours). Spectral clustering divides his social network into three main groups: sports partners, neighbors, and family members. Centrality calculation shows that the betweenness centrality is high in the sports partner group, indicating that it plays an important bridge role in group interaction. Dynamic tracking for a month shows that his social activities are regular, with 2-3 group activities per week, and social intensity and frequency remain stable. Through feature fusion analysis, the old man's social activity score is 85 points (out of 100 points), showing good social participation and interaction quality.

[0066] In a specific embodiment, the process of performing step S104 can specifically include the following steps:

[0067] (1) The physiological index data, cognitive assessment data and activity feature data are subjected to multi-dimensional matrix assembly through a tensor construction algorithm, data alignment is performed in combination with feature mapping relationship, and feature tensor data is obtained;

[0068] (2) High-order singular value decomposition is performed on the feature tensor data, principal components are screened through a kernel norm minimization criterion, and decomposition feature data is obtained;

[0069] (3) A covariance matrix is constructed for the decomposition feature data, feature correlation is quantitatively calculated through subspace analysis, and correlation data is obtained;

[0070] (4) Key feature combinations are extracted from the correlation data, feature structure is reconstructed through non-negative matrix decomposition, and optimization is performed in combination with sparse constraints, and reconstruction data is obtained;

[0071] (5) Multi-level dynamic threshold is set for the reconstruction data, feature classification is performed through ordering statistics and hierarchical clustering, and classified feature data is obtained;

[0072] (6) The classified feature data is compared and analyzed with a group baseline, feature combination optimization is performed through weighted fusion processing, and aging feature data is obtained.

[0073] Specifically, the three types of data (physiological index data, cognitive assessment data and activity feature data) are organized into a three-dimensional tensor structure through a tensor construction algorithm. The three dimensions of the tensor correspond to individuals, feature types and time series, respectively. The feature mapping relationship establishes the correspondence between different types of data. Through time marking, data of different sampling frequencies are mapped to a unified time axis, forming structured feature tensor data. High-order singular value decomposition is performed on the feature tensor data, and the multi-dimensional data is decomposed into a core tensor and a factor matrix. By setting the kernel norm minimization criterion, the principal components that best represent the data are screened out, and noise and redundant information are removed, to obtain the simplified decomposition feature data.

[0074] A covariance matrix is constructed for the decomposition feature data, and the correlation coefficients between different features are calculated. Subspace analysis finds the main feature combination direction through eigenvalue decomposition, quantitatively calculates the feature correlation, and obtains correlation data. When extracting key feature combinations from the correlation data, a non-negative matrix decomposition method is used for feature reconstruction. Given the original data matrix X ∈ R m×n (m is the number of samples, n is the number of features), two non-negative matrices W ∈ R m×k and H ∈ R k×n (k is the number of latent features) are needed, such that:

[0075] X≈WH+E

[0076] Where W represents the basis matrix, H represents the coefficient matrix, and E is the error term. By introducing the sparsity constraint of the L1 norm:

[0077] min||X-WH|| 2 F +λ||W||1+μ||H||1

[0078] Among them, ||·|| F Let ||·||1| represent the Frobenius norm, ||·||1 represent the L1 norm, and λ and μ be regularization parameters. This yields the reconstructed data. Multi-level dynamic thresholds are set for the reconstructed data, and the threshold boundaries are determined based on the data distribution characteristics. The boundary points of each level are determined through ranking statistics, and features are hierarchically grouped using a hierarchical clustering algorithm to obtain hierarchical feature data. The hierarchical feature data is then compared and analyzed with pre-established population baseline data. Through weighted fusion, combining the importance of different features, the feature combinations are optimized and integrated to finally obtain aging feature data.

[0079] For example, an aging characteristic analysis was conducted on a group of 75 elderly individuals in a medical institution. First, the data on each individual's physiological indicators (15 items), cognitive assessments (8 items), and activity characteristics (10 items) were constructed into a 75×33×24 three-dimensional tensor (24 representing hourly time points). After higher-order singular value decomposition, principal components with an energy percentage exceeding 90% were selected, reducing the data dimension to 75×20×24. Correlation analysis revealed a correlation coefficient of -0.72 between blood pressure and cognitive reaction speed, indicating a slight decline in cognitive function when blood pressure rises. Through non-negative matrix factorization, the original 33-dimensional features were reconstructed into 12 key feature combinations. Dynamic thresholds were set to categorize the samples into three levels: mild aging (25 individuals), moderate aging (35 individuals), and severe aging (15 individuals). Finally, a weighted fusion (physiological indicators weight 0.4, cognitive assessment weight 0.35, activity characteristics weight 0.25) was used to derive an aging characteristic score for each individual.

[0080] In one specific embodiment, the process of executing step S105 may specifically include the following steps:

[0081] (1) The aging characteristic data are stratified according to different age groups, the distribution law of the characteristics is analyzed by probability density estimation, and the discrete characteristics are processed into continuous features by combining mathematical statistics principles to obtain the characteristic distribution data.

[0082] (2) Standard score calculation is performed on each dimension index in the feature distribution data, the importance of the index is quantified through entropy weight analysis, the weight vector is initialized and configured combining multi-objective programming theory, and weight coefficient data is obtained;

[0083] (3) A multi-level scoring system is constructed for the weight coefficient data, the feature combination is scored through fuzzy comprehensive evaluation, the scoring rules are corrected combining expert knowledge base, and scoring matrix data is obtained;

[0084] (4) Key evaluation dimensions are extracted from the scoring matrix data, the feature space is reconstructed through principal component analysis, and the feature similarity is calculated combining distance measurement criteria, and feature reorganization data is obtained;

[0085] (5) Comparative analysis is performed on the feature reorganization data and the group baseline, the confidence interval is estimated through Bootstrap sampling, and the abnormal value is corrected combining Bayesian inference, and baseline comparison data is obtained;

[0086] (6) The baseline comparison data is optimized and combined through dynamic programming algorithm, the weight coefficient is dynamically adjusted combining feedback compensation mechanism, and aging quantitative data is obtained through multi-dimensional feature fusion and normalization processing.

[0087] Specifically, the aging feature data is processed in layers. According to age, the data is divided into four levels: youth period (18-44 years old), middle age period (45-59 years old), young old age period (60-74 years old) and high age period (75 years old and above). The distribution rule of the characteristics in each age group is analyzed by kernel density estimation method, and the discrete characteristics (such as social frequency) are continuously processed by spline interpolation method to obtain continuous feature distribution data. The standard score conversion is performed on each index in the feature distribution data, and the indexes with different dimensions are unified into the standard normal distribution interval. Entropy weight analysis calculates the information entropy of each index to evaluate its contribution to the overall evaluation. Combining multi-objective programming theory, the mutual influence between indexes is considered, the initial value of weight is set, and weight coefficient data is generated.

[0088] A multi-level scoring system is constructed for the weight coefficient data, and fuzzy comprehensive evaluation method is used to quantitatively score the feature combination. In fuzzy comprehensive evaluation, let the evaluation object set U = u1, u2,..., u n , the evaluation index set V = v1, v2,..., v m , and the comment set E = e1, e2,..., e k , then the fuzzy evaluation matrix R = (r ij ) n×k , where r ij represents the membership degree of the i-th evaluation object to the j-th comment. The evaluation result B is calculated by the following formula:

[0089]

[0090] wherein A is a weight vector (a1, a2,..., a n ), denotes a fuzzy synthetic operator, b k denotes the final evaluation result of the kth comment. Adjust the scoring rules in combination with the expert knowledge base to finally obtain the scoring matrix data.

[0091] Extract the key evaluation dimensions from the scoring matrix data through principal component analysis, and reconstruct the feature space. Calculate the similarity between features through various distance measurement methods such as Euclidean distance and Manhattan distance, and form the feature reorganization data. Compare the feature reorganization data with the group baseline, use the Bootstrap sampling method to generate a statistical distribution by repeatedly sampling 1000 times, and calculate the 95% confidence interval. For outliers beyond the confidence interval, use Bayesian inference to correct and obtain baseline comparison data.

[0092] Finally, through dynamic programming algorithm, the baseline comparison data is optimized and combined. The feedback compensation mechanism dynamically adjusts the weight coefficient according to the scoring result, and through normalization processing, the final score is mapped to the 0-100 score interval to obtain the aging quantification data.

[0093] For example: A medical research project conducts aging quantification assessment on 500 subjects of different age groups. First, stratify by age: 150 people in the 18-44 age group, 150 people in the 45-59 age group, 120 people in the 60-74 age group, and 80 people in the 75 age group or above. Through kernel density estimation analysis, it is found that physiological indicators (such as blood pressure, heart rate) in different age groups show different distribution characteristics: the standard deviation of the young group is small (blood pressure standard deviation ± 5 mmHg), while the dispersion of the elderly group is large (blood pressure standard deviation ± 12 mmHg). Standard score calculation unifies each indicator to a distribution with a mean of 0 and a standard deviation of 1. Entropy weight analysis shows that the weight of physiological indicators is 0.4, the weight of cognitive function is 0.35, and the weight of social activity is 0.25. In fuzzy comprehensive evaluation, the scoring levels are divided into excellent (90-100 points), good (80-89 points), general (70-79 points), poor (60-69 points), and poor (60 points and below) five levels. Principal component analysis retains the first five principal components that explain 85% of the total variance. The normal value range determined by Bootstrap sampling is: blood pressure index 90-110, cognitive function index 85-115, social activity index 80-120. Finally, through feature fusion, the average score of the 18-44 age group is 88.5, the average score of the 45-59 age group is 82.3, the average score of the 60-74 age group is 76.8, and the average score of the 75 age group or above is 70.2.

[0094] In a specific embodiment, the process of performing step S106 can specifically include the following steps:

[0095] (1) Prioritize each dimension index of the aging quantification data through analytic hierarchy process, construct an intervention target system combined with target decomposition theory, quantify the correlation between targets through coupling degree analysis, and obtain target decomposition data;

[0096] (2) Multi-objective dynamic programming of target decomposition data, allocation and calculation of intervention resources through convex optimization algorithm, screening of strategy space combined with constraint conditions, and obtaining of initial intervention data;

[0097] (3) Constructing a feedback evaluation matrix for the initial intervention data, dynamically evaluating the intervention effect through grey correlation analysis, analyzing the effect trend combined with time sequence feature extraction, and obtaining effect evaluation data;

[0098] (4) Extracting key performance indicators from the effect evaluation data, quantitatively calculating the intervention efficiency through data envelopment analysis, verifying the effect significance combined with difference test, and obtaining efficiency analysis data;

[0099] (5) Multi-dimensional parameter sensitivity analysis of efficiency analysis data, parameter combination optimization through genetic algorithm, exploration of parameter space combined with local search strategy, and obtaining of parameter optimization data;

[0100] (6) Strategy optimization of parameter optimization data through reinforcement learning algorithm, dynamic adjustment of intervention scheme combined with adaptive control theory, comprehensive evaluation of the scheme through multi-criteria decision making, and obtaining of the optimized intervention scheme.

[0101] Specifically, the aging quantification data is analyzed hierarchically, and the evaluation indicators are divided into three dimensions: physiological level (including blood pressure, heart rate, blood glucose, and other basic indicators), cognitive level (including memory, attention, reaction speed, and other cognitive functions), and social activity level (including social frequency, activity intensity, etc.). The priority of each indicator is determined through hierarchical analysis: physiological level accounts for 40%, cognitive level accounts for 35%, and social activity level accounts for 25%. The goal decomposition theory decomposes the overall intervention goal into specific and executable sub-goals, such as decomposing the blood pressure control goal into diet adjustment, exercise planning, and medication guidance. Coupling degree analysis calculates the degree of mutual influence between different goals, such as the positive correlation between exercise intervention and cognitive function improvement, and finally forms the goal decomposition data. The multi-objective dynamic programming method is used to optimize the goal decomposition data, considering various resource constraints such as time, manpower, and material resources. The convex optimization algorithm optimally allocates intervention resources, such as weekly exercise duration allocation and cognitive training frequency arrangement. The constraints include individual physical limitations, upper limit of time investment, medical resource accessibility, etc. Through linear programming, the feasible strategy combination that meets all constraints is screened out to form the initial intervention data.

[0102] A feedback evaluation matrix is constructed for the initial intervention data, including short-term effect indicators (such as immediate reaction ability improvement) and long-term effect indicators (such as physiological indicator stability). Grey correlation analysis evaluates the effectiveness of each intervention measure by calculating the correlation between intervention measures and effect indicators. Time series feature extraction analyzes the time evolution of intervention effects, including immediate effect, cumulative effect, and sustainability, to obtain effect evaluation data. Key performance indicators are extracted from the effect evaluation data, including improvement amplitude, improvement speed, and stability. Data envelopment analysis calculates the input-output ratio of each intervention measure to evaluate intervention efficiency. The statistical significance of intervention effects is verified through difference tests (such as paired t-tests), and comprehensive performance analysis data is formed.

[0103] Multi-dimensional parameter sensitivity analysis is performed on the performance analysis data to investigate the influence of each intervention parameter (such as exercise intensity, training duration, intervention frequency, etc.) on the effect. Genetic algorithm optimizes parameter combinations through operations such as crossover and mutation, and local search strategy conducts fine exploration near the optimal solution to obtain parameter optimization data. Finally, the parameter optimization data is input into the reinforcement learning algorithm, which optimizes the intervention strategy through repeated trials and reward mechanisms. Adaptive control theory dynamically adjusts intervention parameters based on individual feedback, and multi-criteria decision-making considers multiple dimensions such as effect, cost, and sustainability to form the final intervention optimization scheme.

[0104] For example, a medical institution conducts a three-month intervention optimization for a 68-year-old person. The analytic hierarchy shows that the old person needs to focus on cognitive function intervention (weight increased to 45%), followed by physiological indicators (35%), and social activities (20%). The target decomposition refines cognitive function improvement into specific tasks such as attention training (2 times a day, 20 minutes each) and memory exercises (3 times a week, 30 minutes each). The multi-objective planning takes into account the old person's constraint of not more than 2 hours of available time per day, and arranges a gradient progression training plan: the first month focuses on low-intensity cognitive training (total duration 60 minutes / day), the second month increases social activities (cognitive training 45 minutes + social activities 45 minutes / day), and the third month adds moderate exercise (cognitive training 40 minutes + social activities 40 minutes + exercise 30 minutes / day). The grey correlation analysis shows that the correlation between cognitive training and attention improvement is 0.85, with the highest correlation. The performance analysis shows that after three months, the attention test score is increased by 25%, the memory is increased by 20%, and the social activity frequency is increased by 35%. Parameter optimization finds that the best time for cognitive training is from 9 to 11 am, the training duration is appropriate for 20-25 minutes, and the frequency of 2 times a day is the best. Reinforcement learning develops a personalized long-term intervention plan based on these findings, including fixed training time, progressive difficulty setting, and flexible activity combination.

[0105] The clinical aging comprehensive evaluation method in the embodiments of the present application is described above, and the clinical aging comprehensive evaluation system in the embodiments of the present application is described below. Please refer to Figure 3 An embodiment of the clinical aging comprehensive evaluation system in the embodiments of the present application includes:

[0106] The acquisition module is configured to collect physiological indicator data, and obtain physiological indicator data through data denoising and normalization processing.

[0107] The compensation module is configured to perform period testing on cognitive function based on the physiological indicator data, perform compensation processing according to the rhythm curve, and obtain cognitive evaluation data through data fusion analysis.

[0108] The analysis module is configured to track and record behavior characteristics according to the physiological indicator data and the cognitive evaluation data, and obtain activity characteristic data through social frequency calculation and network structure analysis.

[0109] The decomposition module is configured to perform matrix construction and data decomposition on the physiological indicator data, the cognitive evaluation data, and the activity characteristic data, and obtain aging characteristic data through threshold calculation and data reconstruction.

[0110] The recombination module is configured to perform score calculation and feature recombination on the aging characteristic data, and obtain aging quantitative data through baseline comparison and weight adjustment.

[0111] an adjustment module configured to perform target decomposition on the intervention scheme according to the aging quantification data, to obtain an optimized intervention scheme through effect evaluation and parameter adjustment.

[0112] Through the cooperation of the above components, through data denoising and normalization processing, the interference factors in the physiological index data are effectively eliminated, and the quality and reliability of the original data are improved; the cognitive function period test based on the rhythm curve overcomes the defects of ignoring the influence of human physiological rhythm in the traditional test method, and realizes more scientific cognitive ability evaluation; the tracking record and network structure analysis of behavior characteristics break through the limitation of traditional evaluation methods which only focus on single behavior indicators, and realize the comprehensive grasp of social activities and behavior patterns; the application of matrix construction and data decomposition technology solves the problem of unified processing of multi-source heterogeneous data, and realizes the effective integration of physiological, cognitive and behavior data; the process of score calculation and feature reorganization overcomes the subjective shortcomings of traditional scoring methods, and establishes an objective and quantitative evaluation standard; the intervention scheme based on aging quantification data realizes the seamless connection from evaluation to intervention, and improves the accuracy of clinical aging comprehensive evaluation.

[0113] The application also provides a computer readable storage medium, which can be a non-volatile computer readable storage medium, and can also be a volatile computer readable storage medium, and the computer readable storage medium stores instructions, when the instructions run on a computer, the computer executes the steps of the clinical aging comprehensive evaluation method.

[0114] The above-described and above-mentioned embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A comprehensive clinical assessment method for aging, characterized in that, The comprehensive clinical aging assessment method includes: Physiological indicators are collected, and the data are obtained through data denoising and normalization. Based on the aforementioned physiological indicator data, cognitive function is tested over time. Compensation is performed according to the rhythm curve, and cognitive assessment data is obtained through data fusion analysis. This includes: associating and pairing the physiological indicator data with timestamps; extracting physiological fluctuation patterns using a diurnal rhythm analysis algorithm to obtain rhythm baseline data; establishing a multi-dimensional cognitive test matrix based on the rhythm baseline data; obtaining test sequence data by classifying task difficulty and dividing time windows, combined with preset rhythm curves and attention fluctuation curves; setting dynamic response thresholds for the test sequence data; quantifying test results through reaction time statistics and error pattern analysis to obtain raw cognitive data; extracting temporal features from the raw cognitive data; separating cognitive fluctuation patterns using wavelet packet decomposition; evaluating complexity through entropy calculation to obtain cognitive feature data; performing multi-scale analysis on the cognitive feature data; extracting instantaneous phase information using Hilbert transform; correcting the data using a rhythm compensation algorithm to obtain compensated data; and performing multi-level data fusion processing on the compensated data, integrating features using information entropy weight allocation, and adaptively combining multi-dimensional features to obtain cognitive assessment data. Based on the physiological indicator data and the cognitive assessment data, behavioral characteristics are tracked and recorded, and activity characteristic data are obtained through social frequency calculation and network structure analysis. For the physiological indicator data, the cognitive assessment data, and the activity characteristic data, matrix construction and data decomposition are performed, and aging characteristic data are obtained through threshold calculation and data reconstruction. The aging characteristic data are scored and reorganized, and quantitative aging data are obtained through baseline comparison and weight adjustment. Based on the aging quantification data, the intervention plan is decomposed into objectives, and an optimized intervention plan is obtained through effect evaluation and parameter adjustment.

2. The comprehensive clinical aging assessment method according to claim 1, characterized in that, The process of collecting physiological index data, and obtaining physiological index data through data denoising and normalization, includes: Heart rate, blood pressure, and blood glucose were collected in stages according to different time windows by a multi-channel physiological sensor array. After data pre-screening and time-series completion, the original physiological dataset was obtained. Based on the spectral characteristics of the original physiological dataset, adaptive wavelet decomposition is used to separate high-frequency noise and baseline drift, and empirical mode decomposition is combined to eliminate artifacts, resulting in a denoised dataset. Periodic fluctuation features are extracted from the denoised dataset, and the stability of the data is enhanced by combining weighted moving average and nonlinear smoothing techniques. Trend terms are removed by piecewise regression to obtain stable data. A multi-dimensional feature space is constructed for the stationary data. Data dimensionality reduction and reconstruction are performed by principal component rotation and the maximum information entropy criterion. Combined with Z-score standardization, standardized feature data is obtained. Automatically adjustable dynamic threshold boundaries are set for the standardized feature data, outliers are identified through density clustering and anomaly detection algorithms, and data correction is performed in combination with expert rules to obtain corrected data; The corrected data is grouped using hierarchical clustering, and key features are extracted according to the information gain criterion. After feature combination and data alignment, physiological indicator data are obtained.

3. The comprehensive clinical aging assessment method according to claim 1, characterized in that, Based on the physiological indicator data and the cognitive assessment data, behavioral characteristics are tracked and recorded. Activity characteristic data is obtained through social frequency calculation and network structure analysis, including: The physiological indicator data and the cognitive assessment data are time-aligned, and a behavioral time-series benchmark is constructed through multi-source data correlation analysis to obtain benchmark feature data. Multi-dimensional behavior monitoring indicators are set for the benchmark feature data, and daily activities are classified and statistically analyzed using a behavior pattern recognition algorithm to obtain behavior statistics. Social interaction information is extracted from the behavioral statistics, and the interaction intensity and frequency are quantified and calculated through social network topology analysis to obtain social feature data; The social feature data is expanded in a spatiotemporal dimension, and the social group structure is divided by a spectral clustering algorithm. Features are extracted by combining centrality calculation to obtain network structure data. The network structure data is analyzed using graph theory to mine social relationships, and then dynamically tracked using community evolution characteristics to obtain social dynamic data. The social dynamic data is combined with multi-dimensional features, and through time-series correlation analysis and feature fusion processing, combined with the evolution law of behavior patterns, activity feature data is obtained.

4. The comprehensive clinical aging assessment method according to claim 1, characterized in that, The process involves matrix construction and data decomposition of the physiological indicator data, cognitive assessment data, and activity characteristic data. Through threshold calculation and data reconstruction, aging characteristic data is obtained, including: The physiological indicator data, the cognitive assessment data, and the activity feature data are assembled into a multidimensional matrix using a tensor construction algorithm, and the data is aligned by combining feature mapping relationships to obtain feature tensor data. The feature tensor data is subjected to high-order singular value decomposition, and the principal components are screened by the nuclear norm minimization criterion to obtain the decomposed feature data; A covariance matrix is ​​constructed for the decomposed feature data, and the feature correlation is quantitatively calculated through subspace analysis to obtain correlation data. Key feature combinations are extracted from the correlation data, the feature structure is reconstructed through nonnegative matrix factorization, and optimized by combining sparse constraints to obtain reconstructed data. Multi-level dynamic thresholds are set for the reconstructed data, and feature classification is performed through sorting statistics and hierarchical clustering to obtain hierarchical feature data; The hierarchical feature data is compared and analyzed with the population baseline, and the features are combined and optimized through weighted fusion processing to obtain aging feature data.

5. The comprehensive clinical aging assessment method according to claim 1, characterized in that, The process of scoring and reorganizing the aging characteristic data, and obtaining quantitative aging data through baseline comparison and weight adjustment, includes: The aging characteristic data is stratified according to different age groups, the distribution pattern of the characteristics is analyzed by probability density estimation, and the discrete characteristics are processed into continuous data by combining mathematical statistics principles to obtain the characteristic distribution data. Standard scores are calculated for each dimension of the feature distribution data. The importance of the indicators is quantified by entropy weight analysis. The weight vector is initialized and configured by combining multi-objective programming theory to obtain weight coefficient data. A multi-level scoring system is constructed based on the weight coefficient data. The feature combination is scored by fuzzy comprehensive evaluation, and the scoring rules are corrected by combining the expert knowledge base to obtain the scoring matrix data. Key evaluation dimensions are extracted from the scoring matrix data, the feature space is reconstructed through principal component analysis, and the feature similarity is calculated by combining the distance metric criterion to obtain the reconstructed feature data. The reconstructed feature data is compared with the population baseline. Confidence intervals are estimated using Bootstrap sampling, and outliers are corrected using Bayesian inference to obtain baseline comparison data. The baseline comparison data is used to optimize and combine features through a dynamic programming algorithm, and the weight coefficients are dynamically adjusted by a feedback compensation mechanism. Through multi-dimensional feature fusion and normalization, aging quantification data is obtained.

6. The comprehensive clinical aging assessment method according to claim 1, characterized in that, The step of decomposing the intervention plan based on the aging quantification data, and obtaining an optimized intervention plan through effect evaluation and parameter adjustment includes: The aging quantification data is used to prioritize the indicators of each dimension through hierarchical analysis, and an intervention target system is constructed by combining the target decomposition theory. The correlation between targets is quantified through coupling degree analysis to obtain target decomposition data. Multi-objective dynamic programming is performed on the target decomposition data, intervention resources are allocated and calculated using a convex optimization algorithm, and the strategy space is filtered in combination with constraints to obtain initial intervention data; A feedback evaluation matrix is ​​constructed based on the initial intervention data. The intervention effect is dynamically evaluated through grey relational analysis, and the effect trend is analyzed by combining time series feature extraction to obtain effect evaluation data. Key performance indicators are extracted from the effect evaluation data, the intervention efficiency is quantified by data envelopment analysis, and the significance of the effect is verified by difference test to obtain efficacy analysis data. Multidimensional parameter sensitivity analysis is performed on the performance analysis data, and the parameter combination is optimized by using a genetic algorithm. The parameter space is explored by combining a local search strategy to obtain the optimized parameter data. The parameter optimization data is used to optimize the strategy through reinforcement learning algorithm, and the intervention plan is dynamically adjusted by combining adaptive control theory. The plan is then comprehensively evaluated through multi-criteria decision-making to obtain the optimal intervention plan.

7. A comprehensive clinical aging assessment system for implementing the comprehensive clinical aging assessment method as described in any one of claims 1 to 6, characterized in that, The comprehensive clinical aging assessment system includes: The data acquisition module is used to collect physiological indicators and obtain physiological indicator data through data denoising and normalization. The compensation module is used to perform time-segment testing of cognitive function based on the physiological indicator data, perform compensation processing according to the rhythm curve, and obtain cognitive assessment data through data fusion analysis. This includes: associating and pairing the physiological indicator data with timestamps, extracting physiological fluctuation patterns through a diurnal rhythm analysis algorithm to obtain rhythm baseline data; establishing a multi-dimensional cognitive test matrix based on the rhythm baseline data, dividing it into task difficulty levels and time windows, and combining it with preset rhythm curves and attention fluctuation curves to obtain test sequence data; setting dynamic response thresholds for the test sequence data, quantifying test results through reaction time statistics and error pattern analysis to obtain raw cognitive data; extracting temporal features from the raw cognitive data, separating cognitive fluctuation patterns using wavelet packet decomposition, and evaluating complexity through entropy calculation to obtain cognitive feature data; performing multi-scale analysis on the cognitive feature data, extracting instantaneous phase information through Hilbert transform, and correcting the data using a rhythm compensation algorithm to obtain compensated data; and performing multi-level data fusion processing on the compensated data, integrating features through information entropy weight allocation, and adaptively combining multi-dimensional features to obtain cognitive assessment data. The analysis module is used to track and record behavioral characteristics based on the physiological indicator data and the cognitive assessment data, and obtain activity characteristic data through social frequency calculation and network structure analysis. The decomposition module is used to construct matrices and decompose data for the physiological indicator data, the cognitive assessment data, and the activity characteristic data, and to obtain aging characteristic data through threshold calculation and data reconstruction. The recombination module is used to perform scoring calculations and feature recombination on the aging feature data, and obtain aging quantitative data through baseline comparison and weight adjustment; The adjustment module is used to decompose the intervention plan into objectives based on the aging quantification data, and obtain an optimized intervention plan through effect evaluation and parameter adjustment.

8. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the comprehensive clinical aging assessment method as described in any one of claims 1 to 6.

Images