Rehabilitation prediction method and system for multiple organ function decline of old people based on artificial intelligence

Abstract

The invention relates to the technical field of data processing, and discloses an old-age multi-organ function decline rehabilitation prediction method and system based on artificial intelligence. The method comprises the following steps: screening cardiovascular, respiratory, renal function and liver function indexes of an elderly patient through a multi-source data quality evaluation matrix, carrying out time sequence synchronization according to visceral organ physiology time delay characteristics, adopting a visceral organ function correlation positioning analysis algorithm to identify a collaborative decline mode and calculating 1-NN similarity and Max-NN correlation decline degree, and finally, inputting target patient data into a special prediction model of a corresponding hierarchical group to generate a rehabilitation prediction result. According to the method and the device, the technical problem that an existing senile multi-organ function decline prediction method lacks parameter-free collaborative decline mode recognition and individualized layered prediction is solved.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to an artificial intelligence-based method and system for predicting the rehabilitation of multiple organ dysfunction in the elderly. Background Art

[0002] Existing methods for predicting multi-organ dysfunction in the elderly are primarily based on single-organ indicator analysis, assessing a patient's health status by monitoring changes in the functional parameters of individual organs, such as the cardiovascular, respiratory, renal, or liver. Traditional prediction methods use basic machine learning algorithms such as support vector machines and random forests to analyze collected physiological data, predicting a patient's functional decline by establishing a correlation model between single-organ function and disease risk. These methods typically employ unified parameter settings and standardized analysis processes when processing medical data from elderly patients, relying on large-scale labeled data for model training.

[0003] However, existing technologies ignore the complex synergistic mechanisms among multiple organs, resulting in one-sided prediction results that fail to reflect the true condition of multiple organ function decline in elderly patients. Secondly, there are technical bottlenecks in processing the multi-source heterogeneous features of medical data, especially in the spatiotemporal alignment and cross-modal fusion of real-time physiological signals and historical medical records, which makes it difficult to establish an accurate dynamic assessment model. Thirdly, mainstream machine learning models rely too much on large-scale annotated data, lack generalization capabilities in elderly populations with significant individual differences, and lack clinical interpretability, which limits the application value of prediction results in actual clinical decision-making.

[0004] Based on the above analysis, existing technologies are unable to automatically identify the decline characteristics of different elderly patient groups and accurately stratify them. At the same time, there is a lack of dedicated prediction model construction technology for groups with different decline patterns, resulting in low prediction accuracy and insufficient personalization. Summary of the Invention

[0005] The present application provides an artificial intelligence-based method and system for predicting rehabilitation of multiple organ function decline in the elderly, which is used to solve the technical problem of the lack of parameter-free collaborative decline pattern recognition and individualized hierarchical prediction in existing prediction methods for multiple organ function decline in the elderly.

[0006] In the first aspect, the present application provides an artificial intelligence-based rehabilitation prediction method for multiple organ function decline in the elderly, and the artificial intelligence-based rehabilitation prediction method for multiple organ function decline in the elderly includes: screening and processing the cardiovascular function indicators, respiratory function parameters, renal function metabolic data and liver function biochemical indicators of elderly patients through a multi-source data quality assessment matrix to obtain a multiple organ physiological data set; performing time-series synchronization on the multiple organ physiological data set according to the organ physiological time delay characteristics to obtain a multiple organ collaborative monitoring sequence; performing collaborative decline pattern recognition processing on the multiple organ collaborative monitoring sequence through an organ function association positioning analysis algorithm to obtain the organ function 1-NN similarity and Max-NN correlation decline degree; performing adaptive clustering and grouping processing on elderly patients based on the organ function 1-NN similarity and Max-NN correlation decline degree to obtain a multiple organ collaborative decline pattern stratified group; inputting the target elderly patient data into the prediction model corresponding to the multiple organ collaborative decline pattern stratified group for rehabilitation prediction processing to obtain a multiple organ function decline rehabilitation prediction result.

[0007] Optionally, the cardiovascular function indicators, respiratory function parameters, renal function metabolic data and liver function biochemical indicators of elderly patients are screened and processed through the multi-source data quality assessment matrix to obtain a multi-organ physiological data set, including: quality scoring processing of cardiovascular function indicators, respiratory function parameters, renal function metabolic data and liver function biochemical indicators based on four dimensions of data integrity, time continuity, measurement accuracy and clinical relevance to obtain the data quality score of each organ; setting an integrity threshold of 85%, a continuity threshold of 90% and an accuracy deviation threshold of 5% according to the data quality score of each organ to screen the cardiovascular function indicators, respiratory function parameters, renal function metabolic data and liver function biochemical indicators to obtain an organ data group; performing physiological correlation calculation processing on the organ data group by calculating the Pearson correlation coefficient between different organ indicators to obtain an inter-organ correlation coefficient matrix; setting a correlation threshold of 0.3 based on the inter-organ correlation coefficient matrix to perform data validity confirmation processing on the organ data group that meets the quality standards to obtain a multi-organ physiological data set.

[0008] Optionally, the multi-organ physiological data set is time-synchronized according to the physiological time delay characteristics of the organs to obtain a multi-organ collaborative monitoring sequence, including: performing inter-organ time delay characteristic analysis and processing on the multi-organ physiological data set based on the physiological delay characteristics of the impact of changes in cardiac function on renal function and the physiological delay characteristics of the impact of changes in liver function on the cardiovascular system to obtain an inter-organ timing association mapping table; performing intelligent resampling and alignment processing on the high-frequency sampling of electrocardiogram signals, the intermediate-frequency sampling of blood pressure monitoring, and the low-frequency sampling of biochemical indicators according to the inter-organ timing association mapping table to obtain multi-organ data; performing motion artifact and environmental interference elimination processing on the multi-organ data through wavelet transform combined with adaptive filtering algorithm to obtain a denoised multi-organ physiological indicator sequence; performing time base unified correction processing on the denoised multi-organ physiological indicator sequence based on the inter-organ timing association mapping table to obtain a multi-organ collaborative monitoring sequence.

[0009] Optionally, the collaborative decay pattern recognition processing of the multi-organ collaborative monitoring sequence is performed by the organ function correlation positioning analysis algorithm to obtain the organ function 1-NN similarity and Max-NN correlation decay degree, including: based on cardiovascular function, respiratory function, renal function, and liver function as coordinate axes, the multi-organ collaborative monitoring sequence is mapped to a four-dimensional organ function state space to obtain a set of organ function state points of elderly patients; the Euclidean distance calculation formula is used to calculate the most similar functional state distance between the current patient and the historical patient group on the set of organ function state points of elderly patients to obtain the organ function state point. Function 1-NN similarity; based on the calculation of the cosine value of the angle between the multi-organ function decline vectors, the maximum correlation decline pattern matching processing is performed on the organ function status point set of the elderly patients to obtain the Max-NN correlation decline degree; according to the organ function 1-NN similarity and the Max-NN correlation decline degree, the decline pattern recognition processing is performed on the collaborative relationship between cardiovascular-respiratory, kidney-liver, and heart-kidney organs to obtain an individualized multi-organ function decline trajectory; the individualized multi-organ function decline trajectory is input into the organ function association positioning analysis algorithm to perform collaborative decline feature extraction processing to obtain a multi-organ collaborative decline pattern map.

[0010] Optionally, the cardiovascular-respiratory, renal-liver, and heart-kidney organ synergy relationships are subjected to decline pattern recognition processing based on the organ function 1-NN similarity and Max-NN associated decline degree to obtain an individualized multi-organ function decline trajectory, including: constructing a patient similarity network diagram based on the organ function 1-NN similarity to perform similarity matching processing on the cardiovascular-respiratory organ synergy relationship to obtain a cardiovascular-respiratory synergy decline similarity matrix; calculating the renal-liver function decline vector angle based on the Max-NN associated decline degree to perform association strength evaluation processing on the renal-liver organ synergy relationship to obtain a renal-liver synergy decline correlation strength coefficient; performing comprehensive decline pattern analysis processing on the heart-kidney organ synergy relationship through weighted fusion of the organ function 1-NN similarity and Max-NN associated decline degree to obtain a heart-kidney synergy decline comprehensive evaluation index; performing time series trajectory fitting processing on the cardiovascular-respiratory synergy decline similarity matrix, the renal-liver synergy decline correlation strength coefficient, and the heart-kidney synergy decline comprehensive evaluation index to obtain an individualized multi-organ function decline trajectory.

[0011] Optionally, the adaptive clustering and grouping processing of elderly patients based on the organ function 1-NN similarity and Max-NN associated decay degree is performed to obtain a stratified group of multi-organ coordinated decay patterns, including: constructing a patient similarity network diagram according to the organ function 1-NN similarity to perform similarity association modeling processing on the elderly patient group to obtain a functional similarity connection network between patients; setting a dynamic similarity threshold based on the Max-NN associated decay degree to perform cluster boundary determination processing on the functional similarity connection network between patients to obtain an adaptive clustering demarcation parameter; performing cluster center identification processing on the functional similarity connection network between patients by a density peak detection algorithm to obtain a set of cluster center points of a multi-organ decay pattern; dividing the patient group using the adaptive clustering demarcation parameter and the set of cluster center points of the multi-organ decay pattern to obtain a stratified group of multi-organ coordinated decay patterns.

[0012] Optionally, the target elderly patient data is input into the prediction model corresponding to the multi-organ coordinated decline pattern stratification group for rehabilitation prediction processing to obtain the multi-organ functional decline rehabilitation prediction result, including: inputting the target elderly patient data into the special prediction model of the corresponding group in the multi-organ coordinated decline pattern stratification group for patient group matching processing to obtain the decline pattern stratification group identifier to which the patient belongs; calling the corresponding multi-organ coordinated decline pattern stratification group prediction model according to the decline pattern stratification group identifier to perform model activation processing to obtain the activated state special prediction model; performing short-term and medium-term prediction processing on the target elderly patient data through the multi-scale time convolution network in the activated state special prediction model and long-term temporal feature extraction and processing to obtain a multi-temporal feature vector exclusive to the stratified group; the multi-temporal feature vector exclusive to the stratified group is input into the cross-organ feature fusion network in the prediction model corresponding to the multi-organ collaborative decline pattern stratified group for collaborative decline pattern prediction processing to obtain an individualized multi-organ decline trend prediction value; based on the individualized multi-organ decline trend prediction value, a personalized intervention plan is generated by the rehabilitation suggestion generator in the prediction model corresponding to the multi-organ collaborative decline pattern stratified group to obtain a stratified group exclusive rehabilitation guidance plan; the individualized multi-organ decline trend prediction value and the stratified group exclusive rehabilitation guidance plan are comprehensively outputted to obtain a multi-organ functional decline rehabilitation prediction result.

[0013] In a second aspect, the present application provides an artificial intelligence-based rehabilitation prediction system for elderly patients with multiple organ dysfunction, the artificial intelligence-based rehabilitation prediction system for elderly patients with multiple organ dysfunction comprising:

[0014] The screening module is used to screen and process cardiovascular function indicators, respiratory function parameters, renal function metabolic data, and liver function biochemical indicators of elderly patients using a multi-source data quality assessment matrix to obtain a multi-organ physiological data set;

[0015] a synchronization module, configured to synchronize the multi-organ physiological data sets in time sequence according to the physiological time delay characteristics of the organs to obtain a multi-organ collaborative monitoring sequence;

[0016] an identification module, configured to perform collaborative decay pattern recognition processing on the multi-organ collaborative monitoring sequence by using an organ function correlation positioning analysis algorithm to obtain an organ function 1-NN similarity and a Max-NN correlation decay degree;

[0017] Clustering module, used to perform adaptive clustering and grouping of elderly patients based on 1-NN similarity and Max-NN correlation decline of organ function, to obtain stratified groups with multi-organ coordinated decline patterns;

[0018] The prediction module is used to input the target elderly patient data into the prediction model corresponding to the multi-organ coordinated decline pattern stratified group for rehabilitation prediction processing to obtain the multi-organ function decline rehabilitation prediction results.

[0019] In the third aspect, an artificial intelligence-based rehabilitation prediction device for multiple organ function decline in the elderly is provided, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory to enable the artificial intelligence-based rehabilitation prediction device for multiple organ function decline in the elderly to execute the above-mentioned artificial intelligence-based rehabilitation prediction method for multiple organ function decline in the elderly.

[0020] In a fourth aspect, a computer-readable storage medium is provided, wherein instructions are stored in the computer-readable storage medium, which, when executed on a computer, enables the computer to execute the above-mentioned artificial intelligence-based rehabilitation prediction method for elderly multi-organ function decline.

[0021] In the technical solution provided in this application, the four-dimensional evaluation system of the multi-source data quality assessment matrix is ​​used to solve the technical bottleneck that traditional methods cannot effectively handle the multi-source heterogeneous characteristics of medical data, ensure the data quality and correlation of cardiovascular function indicators, respiratory function parameters, renal function metabolic data and liver function biochemical indicators, and lay a solid foundation for subsequent analysis. The introduction of organ physiological time delay characteristics breaks through the limitation of existing technologies that ignore the synergistic mechanism of multiple organs. By establishing a temporal correlation mapping table between organs, accurate spatiotemporal alignment of real-time physiological signals and historical medical record data is achieved, effectively solving the technical difficulties of cross-modal data fusion. The organ function correlation positioning analysis algorithm is the core innovation point. By calculating the 1-NN similarity and Max-NN correlation decay degree of organ function, it realizes parameter-free multi-organ collaborative decay pattern recognition, overcomes the defects of traditional machine learning models that require large-scale labeled data and manual parameter settings, and significantly improves the adaptability and generalization ability in elderly groups with significant individual differences. The adaptive clustering grouping processing based on 1-NN and Max-NN innovatively solves the technical problem of accurate stratification of patient groups. It automatically identifies stratified groups with multi-organ coordinated decline patterns through the density peak detection algorithm, avoids the problem of insufficient personalization caused by the unified processing of traditional methods, and provides a targeted analysis framework for patients with different decline patterns.

[0022] In the application field of rehabilitation prediction for elderly patients with multiple organ function decline, the parameter-free nature of the organ function correlation positioning analysis algorithm enables it to adapt to the differences in physiological characteristics of different elderly patient groups, avoiding the grouping bias that may be caused by the need for preset parameters in traditional clustering algorithms. It is particularly suitable for medical scenarios where there are significant individual differences in the elderly population. The construction of a dedicated prediction model corresponding to the stratified groups of multi-organ collaborative decline patterns breaks through the technical limitations of the existing technology of using a unified model to treat all patients. By training a dedicated prediction network for each stratified group, truly personalized prediction is achieved, significantly improving prediction accuracy and clinical applicability. The application of multi-scale temporal convolutional networks in rehabilitation prediction processing fully considers the temporal characteristics of changes in the functions of multiple organs in elderly patients. By extracting features at three time scales: short-term, medium-term, and long-term, it accurately captures the decline patterns of different time dimensions, and has stronger prediction accuracy than traditional single time scale analysis methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0024] Figure 1 This is a schematic diagram of an embodiment of a method for predicting rehabilitation of multiple organ dysfunction in the elderly based on artificial intelligence in an embodiment of the present application;

[0025] Figure 2 This is a schematic diagram of an embodiment of an artificial intelligence-based rehabilitation prediction system for elderly multiple organ function decline in an embodiment of the present application;

[0026] Figure 3 It is a schematic block diagram of the structure of an artificial intelligence-based rehabilitation prediction device for elderly multiple organ function decline in an embodiment of the present invention. DETAILED DESCRIPTION

[0027] The embodiments of the present application provide a method and system for predicting the rehabilitation of multiple organ function decline in the elderly based on artificial intelligence. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0028] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In the embodiments of the present application, an embodiment of the artificial intelligence-based method for predicting rehabilitation of multiple organ dysfunction in the elderly includes:

[0029] Step S101: screening and processing cardiovascular function indicators, respiratory function parameters, renal function metabolic data, and liver function biochemical indicators of elderly patients using a multi-source data quality assessment matrix to obtain a multi-organ physiological data set;

[0030] Step S102: Synchronizing the time series of multiple organ physiological data sets according to the physiological time delay characteristics of the organs to obtain a multi-organ collaborative monitoring sequence;

[0031] Step S103: performing collaborative decay pattern recognition processing on the multi-organ collaborative monitoring sequence using an organ function correlation positioning analysis algorithm to obtain the organ function 1-NN similarity and Max-NN correlation decay degree;

[0032] Step S104: Adaptively cluster and group the elderly patients based on the 1-NN similarity of organ function and the Max-NN correlation decline degree to obtain a stratified group of multi-organ coordinated decline patterns;

[0033] Step S105: input the target elderly patient data into the prediction model corresponding to the multi-organ coordinated decline pattern stratified group for rehabilitation prediction processing to obtain the multi-organ function decline rehabilitation prediction result.

[0034] It is understandable that the execution subject of this application can be the artificial intelligence-based elderly multiple organ function decline rehabilitation prediction system, or a terminal or server, which is not limited here. The embodiment of this application is described by taking the server as the execution subject as an example.

[0035] Specifically, the data integrity dimension detects missing parameters such as ejection fraction and cardiac output in cardiovascular function indicators. The temporal continuity dimension verifies the consistency of the time intervals for collecting respiratory function parameters. The measurement accuracy dimension assesses the measurement error of creatinine clearance in renal function metabolic data. The clinical relevance dimension analyzes the strength of the association between liver function biochemical indicators and the patient's pathological status. The quality score is calculated by weighted summation of the data quality score for each organ. Subsequently, a triple screening is performed by setting a completeness threshold, a continuity threshold, and an accuracy deviation threshold to ensure that the data entering the subsequent analysis meets the requirements of multi-organ collaborative analysis.

[0036] The analysis of organ physiological time delay characteristics establishes a temporal correlation mapping table between organs based on the physiological delay patterns of changes in cardiac and renal function and the time differences in the impact of the liver and cardiovascular system. This mapping table records the time delay characteristics of the impact of changes in cardiac function on renal function and the delay pattern of abnormal liver function feedback on the cardiovascular system. Intelligent resampling and alignment processing unifies the sampling frequencies through an interpolation algorithm to address the frequency differences between high-frequency sampling of ECG signals, medium-frequency sampling of blood pressure monitoring, and low-frequency sampling of biochemical indicators. Wavelet transform is combined with an adaptive filtering algorithm to eliminate motion artifacts and environmental interference in the data. Adaptive filtering dynamically adjusts the filtering parameters based on signal characteristics.

[0037] The organ function correlation localization analysis algorithm constructs a four-dimensional organ function state space, using cardiovascular, respiratory, renal, and hepatic functions as coordinate axes to establish a spatial mapping of the patient's functional status. The Euclidean distance calculation formula measures the distance between the current patient and the patient with the most similar functional status in a historical patient population. The 1-NN similarity of organ function is calculated by taking the square root of the sum of the squares of the differences between the four organ function indicators. The cosine value of the angle between the multiple organ function decline vectors is calculated by analyzing the angle between the multiple organ function decline trend vectors of different patients to determine the Max-NN correlation decline degree. A smaller angle indicates more similar decline patterns.

[0038] A patient similarity network was constructed based on the 1-NN similarity of organ function. Each node in the network represented a patient, and the edge weights were determined by the 1-NN similarity. The Max-NN correlation decay was used to set a dynamic similarity threshold, which was automatically adjusted based on the distribution of decay patterns across the patient population. A density peak detection algorithm identified high-density regions within the patient similarity network as cluster centers. The algorithm determined the location of cluster centers by calculating the local density of each node and the distance to the high-density point.

[0039] The target elderly patient data is first matched against stratified groups of multi-organ coordinated decline patterns to determine the specific decline pattern group to which the patient belongs. Based on the group identifier, a dedicated prediction model is invoked, with each stratified group corresponding to a different network architecture and parameter configuration. A multi-scale temporal convolutional network incorporates convolution kernels at three time scales: short-term, medium-term, and long-term, capturing intraday fluctuations, cyclical changes, and long-term decline trends in the patient's physiological indicators. A cross-organ feature fusion network uses an attention mechanism to calculate the association weights between different organs, dynamically adjusting the importance of each organ's indicators in the prediction.

[0040] In a specific embodiment, the process of executing step S101 may specifically include the following steps:

[0041] Based on the four dimensions of data integrity, time continuity, measurement accuracy and clinical relevance, cardiovascular function indicators, respiratory function parameters, renal function metabolic data and liver function biochemical indicators were scored and processed to obtain the data quality score of each organ;

[0042] According to the data quality scores of each organ, the integrity threshold of 85%, the continuity threshold of 90%, and the precision deviation threshold of 5% were set to screen and process cardiovascular function indicators, respiratory function parameters, renal function metabolic data, and liver function biochemical indicators to obtain the organ data group;

[0043] By calculating the Pearson correlation coefficient between different organ indicators, the physiological correlation degree of the organ data group is calculated and processed to obtain the inter-organ correlation coefficient matrix;

[0044] Based on the inter-organ correlation coefficient matrix, a correlation threshold of 0.3 was set to confirm the data validity of the organ data group that met the quality standards to obtain a multi-organ physiological data set.

[0045] Specifically, a multi-source data quality assessment matrix addresses the technical bottlenecks of traditional methods in processing multi-source, heterogeneous medical data. The data integrity dimension detects missing values ​​for cardiovascular function parameters such as ejection fraction and cardiac output and calculates the data integrity percentage for each indicator. The temporal continuity dimension verifies whether there are discontinuities in the time intervals between respiratory function parameter collection. The measurement accuracy dimension assesses the measurement error range for renal function metabolic data, such as creatinine clearance. The clinical relevance dimension analyzes the strength of correlation between liver function biochemical indicators and the patient's actual pathological status. Data quality scores for each organ are calculated using a weighted summation method. The scores of the four dimensions are comprehensively evaluated according to preset weights to form a comprehensive quality score for each organ indicator. The screening process implements a triple filtering mechanism by setting a completeness threshold, a continuity threshold, and a precision deviation threshold. The completeness threshold filters out indicators with severe data missingness, the continuity threshold removes monitoring data with discontinuous time series, and the precision deviation threshold filters out physiological parameters with excessive measurement errors. The organ data set includes cardiovascular, respiratory, renal, and liver function indicators that have been screened using these three thresholds. These indicators meet the basic data quality requirements for subsequent analysis.

[0046] The Pearson correlation coefficient calculation reveals the synergistic relationship between changes in multi-organ function by analyzing the degree of linear correlation between different organ indicators. The calculation process includes correlation analysis between cardiovascular function indicators and respiratory function parameters, assessment of the correlation between renal function metabolic data and liver function biochemical indicators, and measurement of the synergistic relationship between cardiac function and renal function. The inter-organ correlation coefficient matrix records the correlation values ​​between all organ indicators. Each element in the matrix represents the Pearson correlation coefficient value between two specific organ indicators.

[0047] Data validation involves final screening based on a correlation threshold set based on the inter-organ correlation coefficient matrix. This threshold is based on the minimum correlation requirement for multi-organ functional synergy in physiological theory. When the Pearson correlation coefficient between organ indicators falls below the set threshold, these indicators lack physiological correlation and are excluded from the multi-organ physiological dataset. The multi-organ physiological dataset contains all organ function indicators that have undergone quality assessment, threshold screening, and correlation verification. These indicators meet not only the data quality standards but also the physiological relevance required for multi-organ synergistic analysis.

[0048] In a specific embodiment, the process of executing step S102 may specifically include the following steps:

[0049] Based on the physiological delay characteristics of the impact of changes in cardiac function on renal function and the physiological delay characteristics of the impact of changes in liver function on the cardiovascular system, the multi-organ physiological data set is analyzed and processed to obtain an inter-organ temporal association mapping table;

[0050] According to the inter-organ temporal correlation mapping table, the high-frequency sampling of ECG signals, the medium-frequency sampling of blood pressure monitoring, and the low-frequency sampling of biochemical indicators are intelligently resampled and aligned to obtain multi-organ data;

[0051] The multi-organ data were processed to eliminate motion artifacts and environmental interference by wavelet transform combined with adaptive filtering algorithm to obtain a de-noised multi-organ physiological index series.

[0052] Based on the inter-organ temporal association mapping table, the denoised multi-organ physiological indicator sequence is uniformly corrected to obtain a multi-organ collaborative monitoring sequence.

[0053] Specifically, the analysis and processing of inter-organ time delay characteristics addresses the technical bottlenecks of traditional methods in temporally and spatially aligning real-time physiological signals with historical medical records. The physiological delay characteristics of the impact of changes in cardiac function on renal function are based on the physiological mechanism of decreased renal perfusion caused by decreased cardiac output. The delay parameters are determined by analyzing the time interval between changes in cardiac ejection fraction and glomerular filtration rate. The physiological delay characteristics of the impact of changes in liver function on the cardiovascular system establish a delay model by monitoring the time difference between the effects of blood composition changes caused by abnormal liver metabolic function on myocardial contractility. The inter-organ temporal association mapping table records the temporal sequence of organ function changes and the delay time of their mutual influence. The table includes key parameters such as the delay time of the heart's impact on the kidneys and the delay time of the liver's impact on the cardiovascular system. Intelligent resampling alignment processing establishes a unified time base to address the differences in sampling frequencies among different monitoring devices. High-frequency sampling of ECG signals refers to the millisecond-level acquisition of heart rhythm data by ECG monitors, medium-frequency sampling of blood pressure monitors refers to the second-level recording of blood pressure changes by sphygmomanometers, and low-frequency sampling of biochemical parameters refers to the hourly or daily acquisition of blood biochemical parameters by laboratory tests. The resampling algorithm uses interpolation to convert data of varying frequencies to a uniform sampling frequency. This interpolation process takes into account the continuity of physiological signals, ensuring that the resampled data retains the original physiological trends. Multi-organ data includes resampled and aligned time series data for cardiovascular, respiratory, renal, and liver function indicators.

[0054] Wavelet transforms use multi-resolution analysis to decompose physiological signals into distinct frequency components, separating useful physiological information from interference components that require removal. Adaptive filtering algorithms dynamically adjust filtering parameters based on signal characteristics. Motion artifacts refer to signal interference caused by patient movement. Environmental interference includes external factors such as electromagnetic interference and equipment noise. The denoising process first uses wavelet transforms to identify the noise frequency range within the signal. Then, an adaptive filtering algorithm applies appropriate filtering strategies to address different types of interference. The denoised multi-organ physiological indicator series contains pure physiological signal data that has undergone noise removal. Time base unified correction processes time-shift the data for each organ based on the delay parameters recorded in the inter-organ temporal correlation mapping table. This correction process aligns the data from different organs according to physiological time relationships, ensuring that the multi-organ data at the same moment reflect the true physiological state. The multi-organ coordinated monitoring series contains all organ function indicator data that have been time-aligned, denoised, and corrected according to physiological time relationships. Each time point in the series contains synchronized measurements of cardiovascular, respiratory, renal, and hepatic function.

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

[0056] Based on cardiovascular function, respiratory function, renal function, and liver function as coordinate axes, the multi-organ collaborative monitoring sequence is mapped into a four-dimensional organ function state space to obtain a set of organ function state points in elderly patients.

[0057] The Euclidean distance calculation formula is used to calculate the distance between the most similar functional states of the current patient and the historical patient group on the organ function status point set of the elderly patients, and the organ function 1-NN similarity is obtained;

[0058] Based on the calculation of the cosine value of the angle between the vectors of multiple organ function decline, the maximum correlation decline pattern matching process is performed on the organ function status point set of elderly patients to obtain the Max-NN correlation decline degree;

[0059] Based on the 1-NN similarity and Max-NN correlation decay of organ functions, the decay pattern recognition processing of cardiovascular-respiratory, kidney-liver, and heart-kidney synergy relationships was performed to obtain individualized multi-organ function decay trajectories.

[0060] The individualized multi-organ function decline trajectory is input into the organ function correlation positioning analysis algorithm for coordinated decline feature extraction to obtain a multi-organ coordinated decline pattern map.

[0061] Specifically, the four-dimensional organ function state space mapping process addresses the problem of traditional methods neglecting the synergistic mechanisms of multiple organs. Four-dimensional spatial mapping constructs a three-dimensional space using cardiovascular, respiratory, renal, and hepatic function as the four coordinate axes. Each elderly patient's multi-organ function state is represented as a unique data point in this space, with the coordinate values ​​of each data point corresponding to the patient's measurement results for each organ function dimension. The set of organ function state points for elderly patients includes all mapped points in the four-dimensional space for both the historical patient population and the current patient. The spatial distribution of these points reflects the differences and similarities in the multi-organ function states of different patients. Euclidean distance calculation determines the degree of functional similarity between patients by measuring the straight-line distance between two points in four-dimensional space. The calculation process involves calculating the numerical difference between the current patient and each patient in the historical patient population in the four functional dimensions of cardiovascular, respiratory, renal, and hepatic function. The square root of these four differences is then summed to obtain the spatial distance. The 1-NN similarity of organ function is determined by finding the historical patient closest to the current patient. The nearest neighbor represents the individual with the most similar multi-organ function state to the current patient, and the distance value reflects the degree of similarity.

[0062] The cosine value of the angle between the multiple organ function decline vectors is calculated based on the trend vectors of the patient's multiple organ function changes over time. The decline vector is formed by calculating the direction and magnitude of change in the patient's organ function indicators at different time points. The cosine value of the angle is calculated by calculating the cosine of the angle between the current patient's decline vector and the decline vectors of historical patient groups to determine the similarity of the decline pattern. The Max-NN correlation decline degree is determined by finding the historical patient with the smallest angle between the current patient's decline vector and the current patient's decline vector. The smaller the angle, the more similar the decline pattern is, and the decline pattern of that patient becomes the maximum correlation decline reference for the current patient.

[0063] Decline pattern recognition identifies coordinated decline relationships between different organs by comprehensively analyzing the 1-NN similarity and Max-NN correlation decline of organ function. Cardiovascular-respiratory synergy is determined by analyzing the synchronization of decreased cardiac ejection fraction and decreased lung function. Renal-hepatic synergy is identified by observing the correlation between changes in glomerular filtration rate and abnormal liver metabolic function. Cardio-renal synergy is established by monitoring the continuity of the effects of changes in cardiac output on renal perfusion. Individualized multi-organ function decline trajectories record the complete path of multi-organ function changes over time for a specific patient. The trajectory includes the decline rate and magnitude of each organ function, as well as the coordinated changes between organs.

[0064] The organ function correlation localization analysis algorithm extracts coordinated decline features by analyzing key characteristic points and change patterns within individualized multi-organ decline trajectories. This feature extraction process includes identifying key information such as inflection points, slope changes, and periodic fluctuations within the trajectories. The multi-organ coordinated decline pattern atlas visualizes these extracted coordinated decline features, displaying the strength of associations between different organs, decline timing, and coordinated evolution patterns. The atlas generates a unique multi-organ decline pattern identifier for each patient.

[0065] In a specific embodiment, the process of performing the decay pattern recognition process on the cardiovascular-respiratory, kidney-liver, and heart-kidney organ synergy relationships based on the organ function 1-NN similarity and Max-NN correlation decay degree may specifically include the following steps:

[0066] Based on the 1-NN similarity of organ function, a patient similarity network diagram was constructed to perform similarity matching on the synergistic relationship between cardiovascular and respiratory organs, and a cardiovascular-respiratory synergistic decline similarity matrix was obtained.

[0067] The kidney-liver function decline vector angle was calculated based on the Max-NN correlation decline degree to evaluate the correlation strength of the kidney-liver synergy relationship and obtain the kidney-liver synergy decline correlation strength coefficient.

[0068] The comprehensive decline pattern analysis of the heart-kidney synergy relationship was performed by weighted fusion of the 1-NN similarity of organ function and the Max-NN correlation decline, and a comprehensive evaluation index of heart-kidney synergy decline was obtained.

[0069] The cardiovascular-respiratory coordinated decline similarity matrix, the renal-hepatic coordinated decline correlation strength coefficient, and the comprehensive assessment index of cardiovascular-renal coordinated decline were fitted with time series trajectories to obtain individualized multi-organ function decline trajectories.

[0070] Specifically, the construction of a patient similarity network graph addresses the problem of traditional methods neglecting the mechanisms of multi-organ synergy. The patient similarity network graph is built based on the 1-NN similarity of organ function. Each node in the network represents an elderly patient, and the edge weights between nodes are determined by the 1-NN similarity of cardiovascular and respiratory function. Patients with higher similarity have greater edge weights. Similarity matching identifies cardiovascular-respiratory synergy by analyzing the synchronization of changes in cardiovascular ejection fraction and respiratory vital capacity. The matching algorithm calculates the similarity values ​​of each pair of patients in the cardiovascular and respiratory function dimensions, forming a cardiovascular-respiratory synergy decline similarity matrix. Each element in the matrix represents the degree of similarity between two patients in their cardiovascular-respiratory synergy decline patterns. The correlation strength assessment process calculates the angle between renal and liver function decline vectors based on the Max-NN correlation decline. The renal function decline vector is formed by analyzing the temporal trends of indicators such as glomerular filtration rate and creatinine clearance, while the liver function decline vector is established by monitoring the decline trajectories of indicators such as transaminase levels and bilirubin metabolism. Vector angle calculation determines the association strength by analyzing the cosine value of the angle between two decline vectors in multidimensional space. The smaller the angle, the stronger the synergy of kidney-liver function decline. The association strength assessment converts the cosine value of the angle into the kidney-liver synergistic decline association strength coefficient, which quantifies the degree of synergy between kidney and liver function decline.

[0071] The comprehensive decline pattern analysis and processing method comprehensively assesses the synergistic relationship between the heart and kidneys through a weighted fusion of the 1-NN similarity and the Max-NN correlation decline. The weighted fusion algorithm sets weight parameters based on the physiological importance of the impact of changes in cardiac function on renal function. The 1-NN similarity reflects the current similarity of cardiac and renal functional states, while the Max-NN correlation decline reflects the historical correlation of cardiac and renal decline patterns. The weighted fusion process linearly combines the two indicators according to preset weights to calculate a comprehensive assessment index of cardiac and renal synergistic decline, which comprehensively reflects the strength and trend of synergistic decline in cardiac and renal function.

[0072] The time series trajectory fitting process uses the cardiovascular-respiratory coordinated decline similarity matrix, the renal-hepatic coordinated decline correlation strength coefficient, and the cardiovascular-renal coordinated decline comprehensive assessment index as input parameters. A time series fitting algorithm is used to construct a personalized multi-organ function decline time trajectory. The fitting algorithm first analyzes the temporal variation patterns of the three input parameters to identify the evolutionary trends of the coordinated relationships between the organs. It then generates a continuous time trajectory curve through interpolation and smoothing. The personalized multi-organ function decline trajectory records the complete path of the coordinated changes in the cardiovascular, respiratory, renal, and hepatic organ functions over time for a specific patient. The trajectory includes the decline rate of each organ function, the time delay of their mutual influence, and the intensity relationship of the coordinated changes.

[0073] In a specific embodiment, the process of executing step S104 may specifically include the following steps:

[0074] Based on the 1-NN similarity of organ functions, a patient similarity network diagram was constructed to perform similarity association modeling on the elderly patient group, and a functional similarity connection network between patients was obtained;

[0075] Based on the Max-NN correlation decay degree, a dynamic similarity threshold is set to determine the cluster boundary of the functional similarity connection network between patients, and the adaptive cluster boundary parameters are obtained;

[0076] The density peak detection algorithm was used to identify the cluster centers of the functional similarity connection network between patients, and the cluster center point set of the multi-organ decline pattern was obtained;

[0077] The adaptive clustering demarcation parameters and the set of cluster center points of the multiple organ decline pattern are used to divide the patient groups to obtain the stratified groups of the multiple organ coordinated decline pattern.

[0078] Specifically, similarity association modeling addresses the technical challenges of traditional methods, which lack individualized difference analysis. A patient similarity network diagram is constructed based on the 1-NN similarity of organ function. Each node in the network represents an elderly patient, and the edges connecting the nodes represent the multi-organ functional similarity relationships between patients. Edge weights are determined by a comprehensive calculation of the 1-NN similarities across four dimensions: cardiovascular, respiratory, renal, and hepatic. The similarity association modeling process analyzes the differences in multi-organ functional status for each pair of patients within the elderly patient population and calculates the similarity values ​​between all pairs of patients to form a complete similarity association matrix. The inter-patient functional similarity connection network connects patient pairs whose similarity exceeds a baseline threshold. The network structure reflects the distribution characteristics and clustering patterns of multi-organ functional status within the elderly patient population. The dynamic similarity threshold is set based on the statistical distribution characteristics of the Max-NN correlation decay. The threshold range is determined by calculating the mean and standard deviation of the Max-NN correlation decay across all patients. The threshold is dynamically adjusted by adding or subtracting multiples of the standard deviation from the mean. Cluster boundary determination involves applying a dynamic similarity threshold to the functional similarity connectivity network between patients. This process selects patient connections that meet the threshold, removes weak connections with low similarity, and retains strong connections with high similarity. Adaptive cluster boundary parameters include dynamically adjusted similarity thresholds and connection strength weights. These parameters are automatically adjusted based on the actual distribution of the patient population, eliminating the manual parameter setting required by traditional methods.

[0079] The density peak detection algorithm identifies cluster centers by analyzing the node density distribution in the functional similarity connection network between patients. The algorithm first calculates the local density of each patient node, which is determined by counting the number and connection strength of neighboring nodes around the patient. The algorithm then calculates the minimum distance from each patient to patients with higher density. The cluster center identification process determines the location of the cluster center by looking for patient nodes with both high local density and large distance values. These nodes represent typical representatives of different multi-organ decline patterns. The set of multi-organ decline pattern cluster center points includes all identified cluster center patients, and each center point corresponds to a specific multi-organ coordinated decline pattern, such as cardiorenal coordinated decline type, respiratory-dominated decline type, and liver and kidney function balanced decline type.

[0080] The patient grouping process uses the adaptive clustering demarcation parameters and the set of multi-organ decline pattern cluster centers as input. The patient's group affiliation is determined by calculating the similarity distance between each patient and each cluster center. The partitioning algorithm assigns each patient to the group corresponding to the cluster center with the highest similarity, while also considering the constraints of the adaptive clustering demarcation parameters to ensure the rationality of the grouping results. The multi-organ decline pattern stratified groups include different patient groups divided according to decline pattern characteristics. Patients within each group have similar multi-organ decline characteristics and development trajectories, and there are significant decline pattern differences between groups.

[0081] In a specific embodiment, the process of executing step S105 may specifically include the following steps:

[0082] The target elderly patient data is input into the dedicated prediction model of the corresponding group in the multi-organ coordinated decline pattern stratification group to perform patient group matching processing, and the decline pattern stratification group identifier to which the patient belongs is obtained;

[0083] According to the patient's decline pattern stratification group identifier, the corresponding multi-organ collaborative decline pattern stratification group prediction model is called to perform model activation processing to obtain a dedicated prediction model in an activated state;

[0084] The multi-scale temporal convolutional network in the activation state-specific prediction model is used to extract short-term, medium-term, and long-term temporal features from the target elderly patient data to obtain multi-temporal feature vectors specific to the stratified groups.

[0085] The multi-time series feature vectors exclusive to the stratified groups are input into the cross-organ feature fusion network in the prediction model corresponding to the multi-organ coordinated decline pattern stratified groups to perform coordinated decline pattern prediction processing, and obtain the individualized multi-organ decline trend prediction value;

[0086] Based on the individualized multi-organ decline trend prediction value, a rehabilitation suggestion generator in the prediction model corresponding to the multi-organ coordinated decline pattern stratified group is used to generate a personalized intervention plan, and a rehabilitation guidance plan exclusive to the stratified group is obtained;

[0087] The individualized multiple organ decline trend prediction value and the stratified group-specific rehabilitation guidance plan are comprehensively output and processed to obtain the multiple organ function decline rehabilitation prediction result.

[0088] Specifically, the patient group matching process addresses the technical issue of traditional methods lacking individualized difference analysis. The target elderly patient data includes cardiovascular function indicators, respiratory function parameters, renal metabolic data, and liver biochemical indicators. The matching process determines patient affiliation by calculating the similarity between the target patient data and the feature vectors of each multi-organ coordinated decline pattern stratified group. A dedicated prediction model is a specialized network model trained for a specific decline pattern group. Each stratified group corresponds to a separate prediction model. The matching algorithm determines the optimal matching group by comparing the distance between the target patient and the central features of each group. The patient's decline pattern stratified group identifier serves as an index for subsequent model invocations and includes group type information and a matching confidence parameter. Model activation uses the patient's decline pattern stratified group identifier to call the corresponding multi-organ coordinated decline pattern stratified group prediction model from the model library. Each prediction model contains network parameters and weight configuration optimized for the specific decline pattern. The activation process includes model parameter loading, network structure initialization, and input interface configuration. The activated dedicated prediction model has the specialized capabilities to process the corresponding patient group data. The internal network layers and connection weights of the model are optimized through training on the historical data of the specific group.

[0089] The multi-scale temporal convolutional network comprises three convolution kernel groups at different time scales: a short-term convolution kernel captures intraday physiological fluctuations, a medium-term convolution kernel identifies periodic functional changes, and a long-term convolution kernel extracts persistent decline trends. The temporal feature extraction process parallelizes convolution operations across three time scales, feeding multi-organ monitoring data from target elderly patients into each of the three convolution kernel groups for feature extraction. Each convolution kernel group outputs a feature vector corresponding to the time scale. A multi-time series feature vector specific to the stratified population is formed by concatenating the feature vectors from the three time scales. This vector captures the functional variations of multiple organs across different time dimensions, reflecting the unique decline patterns of that stratified population.

[0090] The cross-organ feature fusion network calculates the correlation weights between different organ features through an attention mechanism. The fusion process analyzes the collaborative relationships between organs, such as the cardiovascular and respiratory, kidney and liver, and heart and kidney. The collaborative decline pattern prediction process inputs the fused cross-organ features into the prediction layer to calculate the decline trend. The prediction layer generates individualized multi-organ function change trends based on the historical decline patterns of the stratified group. The individualized multi-organ decline trend prediction value includes the future change magnitude, decline rate, and degree of mutual influence of each organ function. The prediction value reflects the patient's personalized development trajectory under the decline pattern of the corresponding stratified group.

[0091] The Rehabilitation Recommendation Generator analyzes patients' specific rehabilitation needs based on their individualized multi-organ decline trend predictions. It includes a built-in library of typical rehabilitation strategy templates and intervention plans for each stratified group. The personalized intervention plan generation process matches the most appropriate intervention measures based on the predicted decline trend, including medication recommendations, rehabilitation training programs, and lifestyle adjustment guidance. Specific rehabilitation guidance plans tailored to the specific decline characteristics of each stratified group are more targeted and applicable.

[0092] Comprehensive output processing integrates individualized predictions of multiple organ function decline trends with customized rehabilitation guidance plans for each stratified group, generating a comprehensive report containing prediction results, risk assessments, and rehabilitation recommendations. The results of the multiple organ function decline rehabilitation prediction are visualized to present the patient's trajectory of multiple organ function changes, risk stratification, and personalized rehabilitation recommendations. The results include specialized predictions and guidance tailored to the characteristics of the stratified group to which the patient belongs.

[0093] The above describes the artificial intelligence-based rehabilitation prediction method for elderly multiple organ function decline in the embodiment of the present application. The following describes the artificial intelligence-based rehabilitation prediction system for elderly multiple organ function decline in the embodiment of the present application. Figure 2 In the embodiment of the present application, an embodiment of the artificial intelligence-based elderly multiple organ function decline rehabilitation prediction system includes:

[0094] The screening module is used to screen and process cardiovascular function indicators, respiratory function parameters, renal function metabolic data, and liver function biochemical indicators of elderly patients using a multi-source data quality assessment matrix to obtain a multi-organ physiological data set;

[0095] a synchronization module, configured to synchronize the multi-organ physiological data sets in time sequence according to the physiological time delay characteristics of the organs to obtain a multi-organ collaborative monitoring sequence;

[0096] an identification module, configured to perform collaborative decay pattern recognition processing on the multi-organ collaborative monitoring sequence by using an organ function correlation positioning analysis algorithm to obtain an organ function 1-NN similarity and a Max-NN correlation decay degree;

[0097] Clustering module, used to perform adaptive clustering and grouping of elderly patients based on 1-NN similarity and Max-NN correlation decline of organ function, to obtain stratified groups with multi-organ coordinated decline patterns;

[0098] The prediction module is used to input the target elderly patient data into the prediction model corresponding to the multi-organ coordinated decline pattern stratified group for rehabilitation prediction processing to obtain the multi-organ function decline rehabilitation prediction results.

[0099] above Figure 2 The artificial intelligence-based rehabilitation prediction system for multiple organ function decline in the elderly in an embodiment of the present invention is described in detail from the perspective of modular functional entities. The artificial intelligence-based rehabilitation prediction device for multiple organ function decline in the elderly in an embodiment of the present invention is described in detail from the perspective of hardware processing.

[0100] Reference Figure 3 In an embodiment of the present invention, there is also provided an artificial intelligence-based rehabilitation prediction device for elderly multiple organ function decline. The artificial intelligence-based rehabilitation prediction device for elderly multiple organ function decline can be a server, and its internal structure can be as follows: Figure 3 As shown. The artificial intelligence-based rehabilitation prediction device for multiple organ function decline in the elderly includes a processor, a memory, a display screen, an input device, a network interface and a database connected via a system bus. Among them, the computer-designed processor is used to provide computing and control capabilities. The memory of the artificial intelligence-based rehabilitation prediction device for multiple organ function decline in the elderly includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the artificial intelligence-based rehabilitation prediction device for multiple organ function decline in the elderly is used to store the corresponding data in this embodiment. The network interface of the artificial intelligence-based rehabilitation prediction device for multiple organ function decline in the elderly is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0101] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the artificial intelligence-based elderly multiple organ function decline rehabilitation prediction device to which the solution of the present invention is applied.

[0102] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the steps of the artificial intelligence-based method for predicting rehabilitation of multiple organ function decline in the elderly.

[0103] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0104] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling an artificial intelligence-based elderly multi-organ function decline rehabilitation prediction device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, etc., various media that can store program code.

[0105] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. An artificial intelligence-based method for predicting the rehabilitation of elderly patients with multiple organ dysfunction, characterized in that: The method comprises: The cardiovascular function indicators, respiratory function parameters, renal function metabolic data and liver function biochemical indicators of elderly patients were screened and processed using a multi-source data quality assessment matrix to obtain a multi-organ physiological data set. Performing time-series synchronization on the multi-organ physiological data sets according to the physiological time delay characteristics of the organs to obtain a multi-organ collaborative monitoring sequence; Performing collaborative decay pattern recognition processing on the multi-organ collaborative monitoring sequence by using an organ function correlation positioning analysis algorithm to obtain the organ function 1-NN similarity and Max-NN correlation decay degree; Adaptive clustering and grouping of elderly patients was performed based on 1-NN similarity and Max-NN correlation decline of organ function, and stratified groups with multi-organ coordinated decline patterns were obtained. The target elderly patients' data were input into the prediction model corresponding to the stratified group of the multi-organ coordinated decline pattern for rehabilitation prediction processing to obtain the rehabilitation prediction results of multi-organ functional decline.

2. The artificial intelligence-based rehabilitation prediction method for elderly patients with multiple organ dysfunction according to claim 1, characterized in that: The multi-source data quality assessment matrix is ​​used to screen and process cardiovascular function indicators, respiratory function parameters, renal function metabolic data, and liver function biochemical indicators of elderly patients to obtain a multi-organ physiological data set, including: Based on the four dimensions of data integrity, time continuity, measurement accuracy and clinical relevance, cardiovascular function indicators, respiratory function parameters, renal function metabolic data and liver function biochemical indicators were scored and processed to obtain the data quality score of each organ; According to the data quality scores of each organ, a completeness threshold of 85%, a continuity threshold of 90%, and an accuracy deviation threshold of 5% are set to screen and process cardiovascular function indicators, respiratory function parameters, renal function metabolic data, and liver function biochemical indicators to obtain an organ data group; Performing physiological correlation calculation on the organ data set by calculating the Pearson correlation coefficient between different organ indicators to obtain an inter-organ correlation coefficient matrix; Based on the inter-organ correlation coefficient matrix, a correlation threshold of 0.3 is set to perform data validity confirmation processing on the organ data group that meets the quality standard to obtain a multi-organ physiological data set.

3. The artificial intelligence-based rehabilitation prediction method for elderly patients with multiple organ dysfunction according to claim 1, characterized in that: The step of performing time-series synchronization on the multiple organ physiological data sets according to the physiological time delay characteristics of the organs to obtain a multiple organ collaborative monitoring sequence includes: Based on the physiological delay characteristics of the impact of changes in cardiac function on renal function and the physiological delay characteristics of the impact of changes in liver function on the cardiovascular system, the multi-organ physiological data set is analyzed and processed to obtain an inter-organ temporal association mapping table; According to the inter-organ temporal association mapping table, high-frequency sampling of electrocardiogram signals, medium-frequency sampling of blood pressure monitoring, and low-frequency sampling of biochemical indicators are intelligently resampled and aligned to obtain multi-organ data; The multi-organ data is processed to eliminate motion artifacts and environmental interference by wavelet transform combined with adaptive filtering algorithm to obtain a de-noised multi-organ physiological index sequence; Based on the inter-organ temporal association mapping table, a time base unified correction process is performed on the denoised multi-organ physiological indicator sequence to obtain a multi-organ collaborative monitoring sequence.

4. The artificial intelligence-based rehabilitation prediction method for elderly patients with multiple organ dysfunction according to claim 1, characterized in that: The method of performing collaborative decay pattern recognition processing on the multi-organ collaborative monitoring sequence by using the organ function correlation positioning analysis algorithm to obtain the organ function 1-NN similarity and Max-NN correlation decay degree includes: Performing four-dimensional organ function state space mapping processing on the multi-organ collaborative monitoring sequence based on cardiovascular function, respiratory function, renal function, and liver function as coordinate axes to obtain a set of organ function state points of elderly patients; The Euclidean distance calculation formula is used to calculate the distance between the most similar functional states of the current patient and the historical patient group on the set of organ function status points of the elderly patient to obtain the organ function 1-NN similarity; Based on the calculation of the cosine value of the angle between the multiple organ function decline vectors, a maximum correlation decline pattern matching process is performed on the organ function state point set of the elderly patient to obtain the Max-NN correlation decline degree; According to the organ function 1-NN similarity and Max-NN correlation decay degree, the cardiovascular-respiratory, kidney-liver, and heart-kidney organ synergy relationships are subjected to decay pattern recognition processing to obtain individualized multi-organ function decay trajectories; The individualized multiple organ function decline trajectory is input into the organ function correlation positioning analysis algorithm to perform coordinated decline feature extraction processing to obtain a multiple organ coordinated decline pattern map.

5. The artificial intelligence-based rehabilitation prediction method for elderly patients with multiple organ dysfunction according to claim 4, characterized in that: The decline pattern recognition process is performed on the cardiovascular-respiratory, kidney-liver, and heart-kidney organ synergy relationships based on the organ function 1-NN similarity and Max-NN correlation decline degree to obtain individualized multi-organ function decline trajectories, including: Based on the organ function 1-NN similarity, a patient similarity network diagram is constructed to perform similarity matching processing on the synergistic relationship between cardiovascular and respiratory organs to obtain a cardiovascular-respiratory synergistic decline similarity matrix; Calculating the angle between the kidney and liver function decline vectors based on the Max-NN correlation decline degree to evaluate the correlation strength of the kidney-liver organ synergy relationship and obtain the kidney-liver synergy decline correlation strength coefficient; The comprehensive decline pattern analysis of the heart-kidney synergy relationship is performed by weighted fusion of the organ function 1-NN similarity and Max-NN correlation decline, thereby obtaining a comprehensive evaluation index of heart-kidney synergy decline; The cardiovascular-respiratory coordinated decline similarity matrix, the renal-liver coordinated decline correlation strength coefficient and the cardiac-renal coordinated decline comprehensive evaluation index are subjected to time series trajectory fitting processing to obtain an individualized multi-organ function decline trajectory.

6. The artificial intelligence-based rehabilitation prediction method for elderly patients with multiple organ dysfunction according to claim 1, characterized in that: The adaptive clustering and grouping process of elderly patients based on the organ function 1-NN similarity and Max-NN correlation decline degree is performed to obtain a stratified group of multi-organ coordinated decline patterns, including: Constructing a patient similarity network diagram based on the organ function 1-NN similarity, performing similarity association modeling on the elderly patient group, and obtaining a functional similarity connection network between patients; Setting a dynamic similarity threshold based on the Max-NN correlation decay degree to perform cluster boundary determination processing on the functional similarity connection network between patients to obtain an adaptive cluster boundary parameter; Performing cluster center identification processing on the functional similarity connection network between patients using a density peak detection algorithm to obtain a set of cluster center points of the multi-organ decline pattern; The adaptive clustering demarcation parameters and the set of cluster center points of the multiple organ decline pattern are used to divide the patient groups to obtain the multiple organ coordinated decline pattern stratified groups.

7. The artificial intelligence-based rehabilitation prediction method for elderly patients with multiple organ dysfunction according to claim 1, characterized in that: The target elderly patient data is input into the prediction model corresponding to the multi-organ coordinated decline pattern stratified group for rehabilitation prediction processing to obtain the multi-organ function decline rehabilitation prediction results, including: Inputting the target elderly patient data into a dedicated prediction model for a corresponding group in a multi-organ coordinated decline pattern stratification group to perform patient group matching processing, and obtaining an identifier of the decline pattern stratification group to which the patient belongs; According to the decline pattern stratification group identifier to which the patient belongs, calling the corresponding multi-organ collaborative decline pattern stratification group prediction model to perform model activation processing to obtain a dedicated prediction model in an activated state; Performing short-term, medium-term, and long-term temporal feature extraction on target elderly patient data using a multi-scale temporal convolutional network in the dedicated prediction model of the activation state to obtain a multi-temporal feature vector exclusive to a stratified group; Inputting the stratified group-specific multi-time series feature vectors into the cross-organ feature fusion network in the prediction model corresponding to the multi-organ coordinated decline pattern stratified group to perform coordinated decline pattern prediction processing to obtain an individualized multi-organ decline trend prediction value; Based on the individualized multiple organ decline trend prediction value, a rehabilitation suggestion generator in the prediction model corresponding to the stratified group of the multiple organ coordinated decline pattern is used to generate a personalized intervention plan to obtain a rehabilitation guidance plan exclusive to the stratified group; The individualized multiple organ function decline trend prediction value and the stratified group exclusive rehabilitation guidance program are comprehensively output and processed to obtain the multiple organ function decline rehabilitation prediction result.

8. An artificial intelligence-based rehabilitation prediction system for elderly patients with multiple organ dysfunction, characterized by: For implementing the artificial intelligence-based rehabilitation prediction method for elderly multiple organ dysfunction according to any one of claims 1 to 7, the artificial intelligence-based rehabilitation prediction system for elderly multiple organ dysfunction comprises: The screening module is used to screen and process cardiovascular function indicators, respiratory function parameters, renal function metabolic data, and liver function biochemical indicators of elderly patients using a multi-source data quality assessment matrix to obtain a multi-organ physiological data set; a synchronization module, configured to synchronize the multi-organ physiological data sets in time sequence according to the physiological time delay characteristics of the organs to obtain a multi-organ collaborative monitoring sequence; an identification module, configured to perform collaborative decay pattern recognition processing on the multi-organ collaborative monitoring sequence by using an organ function correlation positioning analysis algorithm to obtain an organ function 1-NN similarity and a Max-NN correlation decay degree; Clustering module, used to perform adaptive clustering and grouping of elderly patients based on 1-NN similarity and Max-NN correlation decline of organ function, to obtain stratified groups with multi-organ coordinated decline patterns; The prediction module is used to input the target elderly patient data into the prediction model corresponding to the multi-organ coordinated decline pattern stratified group for rehabilitation prediction processing to obtain the multi-organ function decline rehabilitation prediction results.

9. An artificial intelligence-based rehabilitation prediction device for elderly patients with multiple organ dysfunction, characterized by: It includes a memory and a processor, the memory stores a computer program that can be run on the processor, and when the processor executes the computer program, it implements the artificial intelligence-based rehabilitation prediction method for multiple organ function decline in the elderly as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the processor executes the artificial intelligence-based rehabilitation prediction method for multiple organ function decline in the elderly according to any one of claims 1 to 7.

Images