Orthopedic postoperative patient physiological data analysis method and system

Abstract

The invention relates to the technical field of physiological data monitoring, in particular to an orthopedic postoperative patient physiological data analysis method and system.The orthopedic postoperative patient physiological data analysis method comprises the following steps that heart rate blood oxygen and limb activity data are collected, a heart rate fluctuation value, a blood oxygen change value and an activity range are extracted, and a multi-dimensional physiological feature sequence is constructed; and calculating a data change range in the sliding window of each dimension and a frequency change value of adjacent windows. According to the method, fine quantitative analysis of heart rate fluctuation, oxyhemoglobin saturation and limb movement amplitude is achieved by dynamically adjusting data time span and calculating indexes in a segmented mode, change characteristics of different recovery stages are adapted, data cleaning accuracy is improved through high-density region statistics and noise elimination, scientific induction of dynamic characteristics is ensured, and the method has good application prospects. In combination with feature extraction and trend analysis of an acute period, a functional reconstruction period and a stable period, comprehensive evaluation of inter-stage features is formed, and based on an offset rate, a prediction deviation value and abnormal signal trajectory labeling, the anomaly identification and real-time tracking capability is enhanced.

Description

Technical Field

[0001] This invention relates to the field of physiological data monitoring technology, and in particular to a method and system for analyzing physiological data of orthopedic postoperative patients. Background Technology

[0002] The field of physiological data monitoring technology encompasses the real-time or periodic acquisition, processing, and analysis of human physiological parameters. Its core lies in acquiring patients' vital sign data through specific monitoring equipment and analyzing and interpreting this data to aid in diagnosis, treatment, and postoperative recovery. This technological field systematically covers all aspects from physiological signal acquisition, signal sensor design and application, data transmission and storage, to data analysis and presentation. In the medical field, physiological data monitoring technology can be applied to monitoring various indicators such as heart rate, blood pressure, blood oxygen saturation, and body temperature, and achieves a comprehensive assessment of health status by combining biomedical engineering and data analysis methods.

[0003] The "Physiological Data Analysis Method for Post-Orthopedic Surgery Patients" refers to a technical approach for collecting and analyzing specific physiological data related to the recovery process of orthopedic surgery patients. This patent addresses the specific recovery needs of post-orthopedic surgery patients, encompassing the collection of multiple physiological parameters such as heart rate, blood oxygenation, skin temperature, and limb movement data. The data is then cleaned, organized, and analyzed using a specific data processing method based on multi-parameter fusion. Specifically, this method employs high-sensitivity sensors to collect target parameters, removes interference through a signal extraction module, eliminates abnormal data points using a data filtering mechanism, and combines time-series analysis and trend prediction models to label and classify the recovery status. This method achieves comprehensive analysis and judgment of key data during the patient's post-operative recovery process through structured processing.

[0004] Existing technologies struggle to flexibly adjust data analysis strategies based on the dynamic characteristics of a patient's recovery process, resulting in insufficient accuracy in capturing dramatic fluctuations in the acute phase or subtle changes in the stable phase. Data cleaning typically relies on a single threshold, limiting its ability to identify outliers and high-density areas, potentially impacting the reliability of data analysis. Stage feature analysis lacks multi-parameter fusion, relying solely on a single indicator, failing to comprehensively reflect the complex changes during recovery. The absence of stage transition analysis based on probability matrices limits the precise quantification of stage feature correlations, leading to delays in assessing recovery paths and key node changes, thus affecting the real-time adjustment of personalized rehabilitation plans. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and to propose a method and system for analyzing physiological data of orthopedic postoperative patients.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a method for analyzing physiological data of orthopedic postoperative patients, comprising the following steps:

[0007] S1: Collect the patient's heart rate, blood oxygen saturation and limb movement sensor data, extract the heart rate fluctuation value, blood oxygen saturation change value and limb movement amplitude respectively, construct a multidimensional physiological state feature sequence, calculate the data change range within the sliding window and the frequency change value of adjacent windows for each dimension feature in the multidimensional physiological state feature sequence, count the feature change amplitude of continuous data segments, dynamically adjust the window time span to calculate the index in segments, and generate segmented physiological dynamic index values;

[0008] S2: Based on the segmented physiological dynamic index values, extract the local density values ​​of segmental heart rate fluctuations and blood oxygen saturation changes, calculate the number of high-density points around the multi-signal points, remove abnormal noise points according to the distribution distance, summarize the dynamic change characteristics of high-density areas, and generate a temporal clustering feature pattern.

[0009] S3: Based on the time-series clustering feature pattern, calculate the maximum amplitude and frequency interval of the peak signal of heart rate fluctuation in the acute phase, statistically analyze the trend change slope and fluctuation range value during the functional reconstruction period, analyze the signal convergence value and periodic change amplitude characteristics during the stable period, and generate a feature parameter set for the recovery phase.

[0010] S4: Based on the set of characteristic parameters of the recovery stage, analyze the correlation changes of characteristic values ​​from the acute phase to the functional reconstruction phase, calculate the key characteristic value transfer sequence from the functional reconstruction phase to the stable phase, set the time correlation characteristics and transfer relationship matrix of the stage nodes, and generate the characteristic transfer probability matrix of the recovery stage.

[0011] S5: Based on the recovery phase feature transition probability matrix, extract the offset rate and prediction deviation value of the heart rate fluctuation value and blood oxygen saturation change in the current time period, identify the signal mutation trajectory points and determine whether they belong to abnormal areas, mark the abnormal distribution area of ​​trajectory points, and generate the stage abnormal trajectory marking results.

[0012] The segmented physiological dynamic index values ​​include the range of data changes within the sliding window, the frequency change value of adjacent windows, and the characteristic change amplitude of continuous data segments. The temporal clustering feature pattern includes local density values, the number of high-density points, and dynamic change characteristics. The recovery stage feature parameter set includes the maximum amplitude and frequency interval of the peak signal of heart rate fluctuation values, the slope and fluctuation range of the trend change during the functional reconstruction period, the signal convergence value and periodic change amplitude characteristics during the stable period. The recovery stage feature transition probability matrix includes the feature value correlation changes from the acute phase to the functional reconstruction period, the key feature value transition sequence from the functional reconstruction period to the stable period, the temporal correlation characteristics of stage nodes, and the transition relationship matrix. The stage abnormal trajectory annotation results include the offset rate and prediction deviation value of heart rate fluctuation values ​​and blood oxygen saturation changes, signal mutation trajectory points, and abnormal distribution areas.

[0013] As a further aspect of the present invention, the step of obtaining the segmented physiological dynamic index values ​​specifically includes:

[0014] S111: Collect the patient's heart rate, blood oxygen saturation, and limb movement sensor data, extract the heart rate fluctuation value, blood oxygen saturation change value, and limb movement amplitude respectively, construct a multidimensional physiological state feature sequence, and for each dimension feature in the multidimensional physiological state feature sequence, segment the data according to the time span within the sliding window, calculate the difference between the maximum and minimum values ​​of the heart rate fluctuation range, blood oxygen saturation change range, and limb movement amplitude within each sliding window, and obtain the range of variation of basic physiological parameters;

[0015] S112: Combining the time span change trend of the sliding window, perform differential calculation on the change range of the basic physiological parameters of the continuous sliding window. For the heart rate fluctuation range, blood oxygen change range and limb activity amplitude, calculate the parameter change amplitude of adjacent sliding windows one by one. Compare the parameter change amplitude after differential with the time interval of the frequency of adjacent windows. Perform dynamic filtering and aggregation statistics based on the gradient amplitude of the fluctuation range to generate sliding window frequency change characteristic parameters.

[0016] S113: Based on the sliding window frequency change characteristic parameters, the feature values ​​of the segmented data are recalculated by dynamically optimizing and adjusting the time span of the sliding window, using the formula:

[0017] ;

[0018] Generate segmented physiological dynamic index values;

[0019] in, Represents segmented physiological dynamic index values. The weighting coefficients represent the sliding window. This represents the current range of variation in basic physiological parameters. This represents the range of changes in the previous window. Weighting coefficients representing the time span, Represents the current window time span. Represents the time span of the previous window. Represents the number of sliding windows. The number of time spans.

[0020] As a further aspect of the present invention, the step of obtaining the temporal clustering feature pattern specifically includes:

[0021] S211: Based on the segmented physiological dynamic index values, extract the local density values ​​of segmental heart rate fluctuations and blood oxygen saturation changes, calculate the number of high-density points for each signal point within a fixed range, and generate local high-density point distribution parameters.

[0022] S212: Based on the local high-density point distribution parameters, calculate the density change rate between multiple signal points according to the distribution distance between signal points, remove signal points whose distribution distance is higher than the threshold, and subtract the density value of the remaining points from their dynamic gradient to obtain the dynamic change parameters of the high-density region.

[0023] S213: Based on the dynamic change parameters of the high-density region, and by summarizing the temporal dynamic characteristics of the signal point distribution, the following formula is used:

[0024] ;

[0025] Generate temporal clustering feature patterns;

[0026] in, Represents temporal clustering feature patterns. This represents the local density value of the i-th point within a high-density region. Represents the horizontal spatial coordinates of the i-th point. Represents the vertical spatial coordinates of the i-th point. The horizontal spatial coordinates representing the center point of the high-density area. The longitudinal spatial coordinates representing the center point of the high-density area Let i be the weight factor for the i-th point. The dynamic gradient value of the density change of the signal points. These are the weighting coefficients for the dynamic gradient of density changes. This represents the number of signal points within a high-density area. The total number of all filter points. , The iterative index is the summation symbol, representing the iteration range of points within the high-density region and the filter point, respectively.

[0027] As a further aspect of the present invention, the step of obtaining the feature parameter set in the recovery phase specifically includes:

[0028] S311: Based on the time-series clustering feature pattern, extract the maximum amplitude and frequency interval of the peak signal of acute heart rate fluctuation value, filter high amplitude signal data, calculate the difference between the amplitude distribution range and frequency interval of the acute peak signal, and generate acute peak feature parameters.

[0029] S312: Based on the acute phase peak characteristic parameters and combined with the time series data of the functional reconstruction phase signal, calculate the slope of the trend change during the functional reconstruction phase, filter high slope change segments by setting a slope change threshold, perform segmented differential processing on the fluctuation range value of the high slope change segment, and merge the fluctuation range and the change slope to generate the trend characteristic parameters of the functional reconstruction phase.

[0030] S313: Based on the trend characteristic parameters of the functional reconstruction period, combined with the signal data of the stable period, analyze the convergence value of the stable period signal and the average amplitude of the periodic changes, using the formula:

[0031] ;

[0032] Generate a set of feature parameters for the recovery phase;

[0033] in, This represents the set of characteristic parameters for the recovery phase. The amplitude of the signal during the steady-state period, Represents the frequency of the signal. Represents the period of the signal. The weighting factor for amplitude, The absolute value of the signal convergence value. This is an adjustment factor for the signal convergence value. The number of amplitude characteristic signals, This represents the total number of convergence signals.

[0034] As a further aspect of the present invention, the step of obtaining the feature transition probability matrix in the recovery phase specifically includes:

[0035] S411: Based on the set of characteristic parameters of the recovery stage, analyze the correlation changes of characteristic values ​​from the acute phase to the functional reconstruction phase, extract the dynamic change range of characteristic values ​​in the acute phase and the transfer trend of characteristic values ​​in the functional reconstruction phase, calculate the correlation transfer ratio of characteristic values ​​in the two phases, and generate phase characteristic value transfer relationship parameters.

[0036] S412: Based on the stage feature value transfer relationship parameters, combined with the feature value sequence from the functional reconstruction period to the stable period, statistically analyze the key feature value transfer sequence from the functional reconstruction period to the stable period, calculate the time correlation features between multiple feature values ​​in the transfer sequence, remove feature sequences with correlation values ​​lower than a preset threshold, and generate time correlation feature parameters.

[0037] S413: Based on the aforementioned time-related feature parameters, construct the transition relationship matrix between stage nodes using the formula:

[0038] ;

[0039] Generate the feature transition probability matrix for the recovery phase;

[0040] in, To recover the transition probability values ​​in the feature transition probability matrix of the recovery stage, For the time-related feature value of the k-th stage, These are the weight parameters for time-related features. The feature distance between stage nodes. The adjustment factor is the feature distance. This represents the total number of stage nodes.

[0041] As a further aspect of the present invention, the steps for obtaining the stage anomaly trajectory annotation results are specifically as follows:

[0042] S511: Based on the recovery phase feature transition probability matrix, extract the offset rate of heart rate fluctuation and blood oxygen saturation change in the current time period. By comparing the offset rate with the preset reference range point by point, calculate the prediction deviation value of the signal. Statistically analyze the signal feature change amplitude and offset trend to generate offset rate and prediction deviation parameters.

[0043] S512: Based on the offset rate and prediction deviation parameters, identify the change amplitude of the signal mutation trajectory points, combine the temporal continuity of the mutation trajectory points and their spatial distribution distance with neighboring trajectory points, determine whether the trajectory points belong to abnormal areas, and generate trajectory point abnormal distribution area parameters by screening and labeling high deviation trajectory points.

[0044] S513: Based on the abnormal distribution area parameters of the trajectory points, construct the labeling rules for the abnormal trajectories of the stage, using the formula:

[0045] ;

[0046] Label abnormal trajectories during the stage and generate the labeling results for abnormal trajectories during the stage.

[0047] in, The abnormal trajectory annotation value represents the stage. This represents the abnormal deviation value of the i-th trajectory point. The weighting factor represents the trajectory point. Represents the coordinates of the trajectory points. The center coordinates of the region representing the abnormal distribution of trajectory points. This represents the total number of trajectory points.

[0048] A physiological data analysis system for orthopedic postoperative patients, the system being used to execute the aforementioned physiological data analysis method for orthopedic postoperative patients, the system comprising:

[0049] The data dynamic monitoring module obtains the range of data changes within the sliding window and the frequency changes of adjacent windows based on heart rate fluctuations, changes in blood oxygen saturation, and limb movement amplitude data. It also statistically analyzes the characteristic changes of continuous data segments, adjusts the time span of the sliding window for segmented calculation, and generates segmented physiological dynamic index values.

[0050] Based on the segmented physiological dynamic index values, the physiological signal clustering analysis module extracts the local density values ​​of heart rate fluctuations and blood oxygen saturation changes, calculates the number of high-density points around multiple signal points, determines the distribution distance of high-density points and removes abnormal noise points, summarizes the dynamic change characteristics of high-density areas, and generates a time-series clustering feature pattern.

[0051] Based on the time-series clustering feature pattern, the recovery phase feature extraction module calculates the maximum amplitude and frequency interval of the peak signal of acute phase heart rate fluctuation, statistically analyzes the slope and fluctuation range of the trend change during the functional reconstruction period, obtains the convergence value and periodic change amplitude features of the signal during the stable period, and generates a set of feature parameters for the recovery phase.

[0052] Based on the recovery stage feature parameter set, the phase transition probability analysis module analyzes the correlation changes of feature values ​​from the acute phase to the functional reconstruction phase, calculates the key feature value transition sequence from the functional reconstruction phase to the stable phase, sets the phase node time correlation features and transition relationship matrix, and generates the recovery stage feature transition probability matrix.

[0053] The abnormal trajectory recognition module extracts heart rate fluctuation values ​​and blood oxygen saturation change offset rates based on the feature transition probability matrix of the recovery phase, calculates prediction deviation values ​​and identifies trajectory points of signal mutations, determines whether the trajectory points belong to abnormal areas and marks abnormal distribution areas, and generates stage abnormal trajectory marking results.

[0054] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0055] This invention achieves refined quantitative analysis of heart rate fluctuations, blood oxygen saturation, and limb movement amplitude by dynamically adjusting the time span and segmented calculation indicators of the data, adapting to the changing characteristics of different recovery stages. Statistical analysis of high-density areas and noise removal improve the accuracy of data cleaning, ensuring the scientific summarization of dynamic change characteristics. Combining dynamic feature extraction and trend analysis of the acute, functional reconstruction, and stable phases, a comprehensive assessment of inter-phase characteristics is formed, clarifying key transfer patterns. Trajectory annotation based on offset rate, prediction deviation value, and abnormal signals enhances the ability to identify and track abnormal situations in real time, supporting hierarchical management and precise intervention in the postoperative rehabilitation process. Attached Figure Description

[0056] Figure 1 This is a schematic diagram of the workflow of the present invention;

[0057] Figure 2 This is a flowchart of the steps for obtaining segmented physiological dynamic index values ​​according to the present invention;

[0058] Figure 3 This is a flowchart illustrating the steps for obtaining temporal clustering feature patterns according to the present invention.

[0059] Figure 4 This is a flowchart of the steps for obtaining the feature parameter set during the recovery phase of this invention.

[0060] Figure 5 This is a flowchart illustrating the steps for obtaining the feature transition probability matrix during the recovery phase of this invention.

[0061] Figure 6 This is a flowchart illustrating the steps for obtaining abnormal trajectory annotation results during the current invention. Detailed Implementation

[0062] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0063] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0064] Example 1

[0065] Please see Figure 1 This invention provides a technical solution: a method for analyzing physiological data of orthopedic postoperative patients, comprising the following steps:

[0066] S1: Collect the patient's heart rate, blood oxygen saturation and limb movement sensor data, extract the heart rate fluctuation value, blood oxygen saturation change value and limb movement amplitude respectively, construct a multidimensional physiological state feature sequence, calculate the data change range within the sliding window and the frequency change value of adjacent windows for each dimension feature in the multidimensional physiological state feature sequence, count the feature change amplitude of continuous data segments, dynamically adjust the window time span to calculate the index in segments, and generate segmented physiological dynamic index values;

[0067] S2: Based on segmented physiological dynamic index values, extract the local density values ​​of segmented heart rate fluctuations and blood oxygen saturation changes, calculate the number of high-density points around multiple signal points, remove abnormal noise points according to distribution distance, summarize the dynamic change characteristics of high-density areas, and generate temporal clustering feature patterns.

[0068] S3: Based on the temporal clustering feature pattern, calculate the maximum amplitude and frequency interval of the peak signal of heart rate fluctuation in the acute phase, statistically analyze the trend change slope and fluctuation range value during the functional reconstruction period, analyze the signal convergence value and periodic change amplitude characteristics during the stable period, and generate a feature parameter set for the recovery phase.

[0069] S4: Based on the feature parameter set of the recovery stage, analyze the correlation changes of feature values ​​from the acute phase to the functional reconstruction phase, calculate the key feature value transfer sequence from the functional reconstruction phase to the stable phase, set the time correlation features and transfer relationship matrix of the stage nodes, and generate the feature transfer probability matrix of the recovery stage.

[0070] S5: Based on the feature transition probability matrix of the recovery phase, extract the offset rate and prediction deviation value of the heart rate fluctuation value and blood oxygen saturation change in the current time period, identify the signal mutation trajectory points and determine whether they belong to the abnormal area, mark the abnormal distribution area of ​​the trajectory points, and generate the stage abnormal trajectory labeling results.

[0071] The segmented physiological dynamic index values ​​include the range of data changes within the sliding window, the frequency changes of adjacent windows, and the characteristic change amplitude of continuous data segments. The temporal clustering feature patterns include local density values, the number of high-density points, and dynamic change characteristics. The recovery stage feature parameter set includes the maximum amplitude and frequency interval of the peak signal of heart rate fluctuation values, the slope and fluctuation range of the trend change during the functional reconstruction period, the convergence value of the signal during the stable period, and the periodic change amplitude characteristics. The recovery stage feature transition probability matrix includes the correlation changes of feature values ​​from the acute phase to the functional reconstruction period, the key feature value transition sequence from the functional reconstruction period to the stable period, the temporal correlation characteristics of stage nodes, and the transition relationship matrix. The stage abnormal trajectory annotation results include the offset rate and prediction deviation value of heart rate fluctuation values ​​and blood oxygen saturation changes, signal mutation trajectory points, and abnormal distribution areas.

[0072] Please see Figure 2 The specific steps for obtaining segmented physiological dynamic index values ​​are as follows:

[0073] S111: Collect the patient's heart rate, blood oxygen saturation, and limb movement sensor data, extract the heart rate fluctuation value, blood oxygen saturation change value, and limb movement amplitude respectively, construct a multidimensional physiological state feature sequence, and for each dimension feature in the multidimensional physiological state feature sequence, segment the data according to the time span within the sliding window, calculate the difference between the maximum and minimum values ​​of the heart rate fluctuation range, blood oxygen saturation change range, and limb movement amplitude within each sliding window, and obtain the range of variation of basic physiological parameters;

[0074] Heart rate fluctuation values ​​are collected from time series data at fixed time intervals to obtain the maximum and minimum heart rate values ​​within each time segment and calculate the fluctuation range. Blood oxygen saturation changes are also collected in segments through a sliding time window, and the difference between the maximum and minimum blood oxygen saturation changes is extracted as the variation range for each segment. Limb activity amplitude is collected from raw data recorded by an accelerometer, and the amplitude of acceleration changes over a continuous time period is calculated as the range of activity amplitude changes. By integrating the above three types of numerical features, the range of heart rate fluctuation, blood oxygen saturation change, and limb activity amplitude change within the time window are generated respectively, thus obtaining the range of changes in basic physiological parameters.

[0075] S112: Combining the time span change trend of the sliding window, perform differential calculation on the range of basic physiological parameters of continuous sliding windows. For the range of heart rate fluctuation, range of blood oxygen change, and range of limb movement, calculate the parameter change amplitude of adjacent sliding windows one by one. Compare the parameter change amplitude after differential with the time interval of adjacent window frequencies. Perform dynamic filtering and aggregation statistics based on the gradient amplitude of the fluctuation range to generate sliding window frequency change characteristic parameters.

[0076] For the range of variation of basic physiological parameters in a continuous sliding window, the parameters of each window are differentially calculated according to the time series. The difference value of the heart rate fluctuation range is obtained by calculating the absolute difference between the current value of the heart rate fluctuation range and the value of the previous window window. The difference value of the blood oxygen saturation fluctuation range is calculated by subtracting the previous window value from the current window value of the blood oxygen saturation fluctuation range. The difference value of limb movement amplitude is calculated according to the difference in acceleration change amplitude between adjacent windows. The range of variation of all difference values ​​is statistically analyzed, and significant change segments in the difference values ​​are screened out. Data segments with obvious fluctuations are screened by setting a threshold for blood oxygen difference values. The difference amplitude is aggregated and statistically analyzed using the gradient method. The three types of change characteristics of heart rate, blood oxygen and limb movement amplitude are integrated to generate sliding window frequency change characteristic parameters.

[0077] S113: Based on the characteristic parameters of the sliding window frequency variation, the feature values ​​of the segmented data are recalculated by dynamically optimizing and adjusting the time span of the sliding window, using the following formula:

[0078] ;

[0079] Generate segmented physiological dynamic index values;

[0080] in, Represents segmented physiological dynamic index values. The weighting coefficients represent the sliding window. This represents the current range of variation in basic physiological parameters. This represents the range of changes in the previous window. Weighting coefficients representing the time span, Represents the current window time span. Represents the time span of the previous window. Represents the number of sliding windows. The number of time spans.

[0081] formula:

[0082] ;

[0083] The advantage of the formula is that by combining the range of variation of basic physiological parameters with the dynamic changes of time span, it optimizes the adjustment method of the time window, which can reflect the sensitivity of the sliding window frequency characteristic parameters to the time span, thereby generating segmented physiological dynamic index values ​​more accurately.

[0084] Detailed explanation of the formula and its calculation derivation:

[0085] Represents segmented physiological dynamic index values. The weight coefficients representing the sliding window are obtained through normalized statistics of the difference magnitudes within the window. This represents the current range of variation in basic physiological parameters, calculated using the maximum and minimum values ​​within the window. This represents the range of variation of the basic physiological parameters from the previous window, and is directly retrieved from the calculated values ​​of the previous window. The weighting coefficients representing the time span are obtained through normalized statistical analysis of the gradient change magnitude over the time span. This represents the current window's time span, calculated directly from the sliding window's time segments. This represents the time span of the previous window, and can also be directly accessed by sliding the window to retrieve the previous time segment. This represents the number of sliding windows, which is determined by counting the number of segments sampled during the sliding process. The number of time spans is calculated by accumulating the time segments and combining the above parameters:

[0086] ;

[0087] The results indicate that the segmented physiological dynamic index value is 1, reflecting the weight distribution and gradient fluctuation level between the sliding window frequency change characteristic parameter and the dynamic adjustment time span, providing support for subsequent optimization of the sliding window parameters.

[0088] Please see Figure 3 The specific steps for obtaining temporal clustering feature patterns are as follows:

[0089] S211: Based on segmented physiological dynamic index values, extract the local density values ​​of segmental heart rate fluctuations and blood oxygen saturation changes, calculate the number of high-density points for each signal point within a fixed range, and generate local high-density point distribution parameters.

[0090] Based on the spatial distribution of signal points and the data distribution trend within a local area, the number of neighboring points of each signal point within a preset range is calculated, and the number of local high-density points is counted one by one. In the calculation of the number of neighboring points, a radius search method is used. Each signal point is set as the center, and the number of all signal points within its coverage area is calculated with a specified radius. The result generates a local density distribution value. On this basis, signal points higher than the local density threshold are selected as local high-density points. Regarding the setting of the density threshold, firstly, the local density values ​​of all signal points are sorted, and the 80% quantile of the sorting result is taken as the threshold benchmark value. Then, high-density points are selected. By counting the number of high-density points within the range of each signal point, local high-density point distribution parameters are generated.

[0091] S212: Based on the distribution parameters of local high-density points, calculate the density change rate between multiple signal points according to the distribution distance between signal points, remove signal points whose distribution distance is higher than the threshold, and subtract the density value of the remaining points from their dynamic gradient to obtain the dynamic change parameters of the high-density region.

[0092] By calculating the Euclidean distance between high-density points, signal points with distances exceeding a set threshold are filtered out as noise points. The threshold is determined by calculating the mean and standard deviation of the distance distribution of all high-density points, and using the mean plus 1.5 times the standard deviation as the distribution distance threshold. All signal points exceeding the threshold are removed as noise points. The remaining signal points are calculated based on their local density values ​​and dynamic density gradients. The dynamic gradient is calculated by differentiating the density values ​​of the signal points, and the result is the absolute difference between the density values ​​of two consecutive signal points. The dynamic density gradient of the signal points is then combined with their spatial coordinates to generate a signal point density gradient distribution table. The remaining signal points in the table are categorized and summarized to obtain the dynamic change parameters of the high-density area.

[0093] S213: Based on the dynamic change parameters of high-density areas, the formula is used by summarizing the temporal dynamic characteristics of signal point distribution:

[0094] ;

[0095] Generate temporal clustering feature patterns;

[0096] in, Represents temporal clustering feature patterns. This represents the local density value of the i-th point within a high-density region. Represents the horizontal spatial coordinates of the i-th point. Represents the vertical spatial coordinates of the i-th point. The horizontal spatial coordinates representing the center point of the high-density area. The longitudinal spatial coordinates representing the center point of the high-density area Let i be the weight factor for the i-th point. The dynamic gradient value of the density change of the signal points. These are the weighting coefficients for the dynamic gradient of density changes. This represents the number of signal points within a high-density area. The total number of all filter points. , The iterative index is the summation symbol, representing the iteration range of points within the high-density region and the filter point, respectively.

[0097] formula:

[0098] ;

[0099] The advantage of the formula is that by combining the local density, dynamic density gradient, and distance weight of the spatial distribution center of the signal points, it can summarize the dynamic characteristics of the signal points in a time series, which effectively improves the accuracy of clustering feature patterns and the ability to identify dynamic features.

[0100] Detailed explanation of the formula and its calculation derivation:

[0101] Represents temporal clustering feature patterns. The local density value of the i-th point within the high-density region is obtained by counting the number of signal points within the local search range. and Let represent the spatial coordinates of the i-th point, which are obtained directly from the spatial data of the signal points. The coordinates of the center point of the high-density region are calculated using the average value of signal points within the region. The density weighting factor for the signal points is calculated by normalizing the signal point density values. The dynamic density gradient value is obtained by calculating the rate of change of the signal point density value using differential calculation. The weighting factors for the dynamic gradient are obtained through dynamic gradient value normalization. These represent the number of signal points in the high-density area and the total number of filtered signal points, respectively.

[0102] Substitute the above parameter values ​​into the formula:

[0103] Step 1, calculate the molecule:

[0104] ;

[0105] Calculate item by item:

[0106] The first item is: ;

[0107] The second item is: ;

[0108] The third item is: ;

[0109] The fourth item is: ;

[0110] The sum of the numerators is: .

[0111] The second step is to calculate the denominator:

[0112] ;

[0113] Third step, calculation results:

[0114] ;

[0115] The results show that the dynamic clustering index of the temporal clustering feature pattern is 0.742, indicating that the distribution of signal points in the high-density region has high dynamic feature consistency, which is related to the step results and represents that the dynamic clustering trend of the signal points has been effectively captured and quantified.

[0116] Please see Figure 4 The specific steps for obtaining the feature parameter set during the recovery phase are as follows:

[0117] S311: Based on the temporal clustering feature pattern, extract the maximum amplitude and frequency interval of the peak signal of acute heart rate fluctuation value, filter high amplitude signal data, calculate the difference between the amplitude distribution range and frequency interval of the acute peak signal, and generate acute peak feature parameters.

[0118] First, the heart rate fluctuation values ​​are segmented and statistically analyzed. The maximum amplitude variation range of each segment and the frequency interval between adjacent signal peaks are calculated using a dynamic window. Signal points with amplitude values ​​greater than the interval threshold are extracted. The characteristics of the signal points are further analyzed. By calculating the difference between the maximum and minimum peak values ​​of each signal segment and combining the gradient change of amplitude distribution, local features are calculated. High amplitude signal points are screened by frequency interval values ​​and amplitude differences. High-frequency peak signals that meet the characteristics of acute phase fluctuations are further screened to generate acute phase peak feature parameters.

[0119] S312: Based on the peak characteristic parameters of the acute phase and combined with the time series data of the functional reconstruction phase signal, calculate the slope of the trend change during the functional reconstruction phase. By setting a slope change threshold, filter high slope change segments. Perform segmented differential processing on the fluctuation range value of the high slope change segment, and merge the fluctuation range and the change slope to generate the trend characteristic parameters of the functional reconstruction phase.

[0120] The slope of the trend change during the functional reconstruction period is calculated by linear fitting. The rate of change of the signal in the high slope segment is statistically analyzed segment by segment. The trend change is segmented by calculating the difference in the fluctuation range. The maximum slope value and average fluctuation range of each segment are extracted. Regions in the signal segment that meet the trend characteristics are screened. The slope fluctuation value is adjusted after the signal of the screened region is differentially processed. All slope segment characteristic parameters are merged by superposition analysis to generate the trend characteristic parameters of the functional reconstruction period.

[0121] S313: Based on the trend characteristic parameters during the functional reconstruction period, combined with the signal data during the stable period, the convergence value and the average amplitude of the periodic changes of the signal during the stable period are analyzed using the following formula:

[0122] ;

[0123] Generate a set of feature parameters for the recovery phase;

[0124] in, This represents the set of characteristic parameters for the recovery phase. The amplitude of the signal during the steady-state period, Represents the frequency of the signal. Represents the period of the signal. The weighting factor for amplitude, The absolute value of the signal convergence value. This is an adjustment factor for the signal convergence value. The number of amplitude characteristic signals, This represents the total number of convergence signals.

[0125] formula:

[0126] ;

[0127] The advantage of the formula is that by dynamically combining the amplitude, frequency, and periodicity of the signal with the signal convergence value, it is possible to quantify the changes in the periodicity of the signal during the recovery phase, thereby improving the accuracy of feature extraction.

[0128] Detailed explanation of the formula and its calculation derivation:

[0129] Suppose that a certain stable period signal segment contains the following characteristic data:

[0130] signal amplitude : signal frequency : Signal period : Weighting factor : signal convergence value : Adjustment coefficient : .

[0131] Calculate the weighted value of the amplitude:

[0132] ;

[0133] ;

[0134] Calculate the weighted sum of the convergent values ​​of the signal:

[0135] ;

[0136] Substitute the parameters into the formula to calculate the characteristic parameters of the recovery phase:

[0137] ;

[0138] The results indicate that the low values ​​of the characteristic parameter set during the recovery phase suggest that the periodic changes of the signal are relatively stable and that the signal has a high correlation with the characteristics of the stable phase signal. This data can be further used to analyze the overall recovery phase characteristics.

[0139] Please see Figure 5 The specific steps for obtaining the feature transition probability matrix during the recovery phase are as follows:

[0140] S411: Based on the feature parameter set of the recovery stage, analyze the correlation changes of feature values ​​from the acute phase to the functional reconstruction phase, extract the dynamic change range of feature values ​​in the acute phase and the transfer trend of feature values ​​in the functional reconstruction phase, calculate the correlation transfer ratio of feature values ​​between the two phases, and generate phase feature value transfer relationship parameters.

[0141] First, feature value change data of each signal point in the acute phase are extracted. The trend of heart rate peak value and its fluctuation range is statistically analyzed segmentally. A sliding window is used to statistically analyze the maximum, minimum, and mean heart rate changes within each window to obtain the variation amplitude of feature values ​​within each window. This is compared with the time series distribution of feature values ​​in the functional reconstruction phase. The slope variation range is dynamically adjusted based on the time series of feature values ​​to extract highly correlated heart rate signal points. The correlation ratio between the feature values ​​of the two phases is calculated using the formula: in, Characteristic values ​​representing acute phase signal points, The associated weight values ​​of signal points during the functional reconstruction period. The total number of features within the sliding window. This refers to the number of signal points in the acute phase. To determine the number of signal points during the functional reconstruction period, the specific calculation process involves dynamically calculating based on the characteristic values ​​within the interval. Signal points whose characteristic value change rate exceeds the threshold are removed, and the correlation of characteristic points is recalculated to obtain the correlation transfer ratio of characteristic values ​​between the two stages, thereby generating stage characteristic value transfer relationship parameters.

[0142] S412: Based on the stage feature value transfer relationship parameters, combined with the feature value sequence from the functional reconstruction period to the stable period, statistically analyze the key feature value transfer sequence from the functional reconstruction period to the stable period, calculate the time correlation features between multiple feature values ​​in the transfer sequence, remove feature sequences with correlation values ​​lower than the preset threshold, and generate time correlation feature parameters.

[0143] First, the signal points during the functional reconstruction period are segmented into time series, and signal segments with significant characteristic fluctuations are extracted. By calculating the signal peak value, trough value, and average rate of change, the dynamic characteristic value range of each signal segment is obtained. These dynamic characteristic values ​​are compared one by one with the signal segments during the stable period. Based on the frequency distribution characteristics of the time series, the similarity between the two-stage signal segments is calculated. The time nodes of characteristic value transition are divided according to the similarity. By accumulating the signal change values ​​within the transition time period, the time correlation coefficient of the characteristic values ​​is calculated. A minimum threshold for the time correlation coefficient is set, and signal segments below the threshold are removed. Highly correlated signal segments in the time series are extracted, and their dynamic characteristic values ​​are segmented to generate time-related characteristic parameters.

[0144] S413: Construct a transition relationship matrix between stage nodes based on time-related feature parameters, using the formula:

[0145] ;

[0146] Generate the feature transition probability matrix for the recovery phase;

[0147] in, To recover the transition probability values ​​in the feature transition probability matrix of the recovery stage, For the time-related feature value of the k-th stage, These are the weight parameters for time-related features. The feature distance between stage nodes. The adjustment factor is the feature distance. This represents the total number of stage nodes.

[0148] formula:

[0149] ;

[0150] The advantage of the formula is that by introducing a combination of time-related feature parameters and weight parameters, and combining them with a dynamic adjustment factor for node feature distance, it can more accurately reflect the feature transfer probability distribution between stage nodes, which helps to refine the correlation of feature changes between stage nodes.

[0151] Detailed explanation of the formula and its calculation derivation:

[0152] Assume the number of nodes in the acute phase is 5, the number of nodes in the functional reconstruction phase is 7, the number of nodes in the stable phase is 3, and the total number of nodes is... Take the acute phase time-related feature value Weight parameters Node distance Adjustment factor .

[0153] The first step is to calculate the numerator of the transition probability value for each node:

[0154] ;

[0155] Substitute the parameters respectively:

[0156] ;

[0157] The calculated molecular result is as follows:

[0158] ;

[0159] The second step is to calculate the denominator:

[0160] ;

[0161] The calculated denominator is:

[0162] ;

[0163] The third step is to calculate the transition probability value:

[0164] ;

[0165] The results indicate that the feature transfer probability of nodes between the acute phase and the functional reconstruction phase is 85%, suggesting a high correlation between the feature values ​​of the two phases. The feature transfer probability matrix of the recovery phase can be further generated through the dynamic adjustment mechanism of this formula.

[0166] Please see Figure 6 The specific steps for obtaining the stage anomaly trajectory annotation results are as follows:

[0167] S511: Based on the feature transition probability matrix of the recovery phase, extract the offset rate of the heart rate fluctuation value and blood oxygen saturation change in the current time period. By comparing the offset rate with the preset reference range point by point, calculate the prediction deviation value of the signal. Statistically analyze the signal feature change amplitude and offset trend to generate offset rate and prediction deviation parameters.

[0168] First, the instantaneous heart rate and short-term trend value are extracted from the monitored heart rate fluctuations. The offset rate is obtained by calculating the difference between the instantaneous heart rate and the short-term trend value. The offset rate can be calculated using the formula... Calculation, where This is the instantaneous value of heart rate, acquired once per minute by a dynamic monitoring device. The short-term trend value can be calculated using the moving average of heart rate data every 5 minutes. The corresponding blood oxygen saturation change deviation rate can be calculated in a similar way. The deviation rate result is further compared with the set reference range, such as a heart rate range of 40-200 bpm and a blood oxygen saturation range of 90%-100%. Points with deviation rates exceeding the set range are filtered out and marked as deviation anomalies. At the same time, the prediction deviation value is calculated by the difference between the actual value and the output value of the prediction model, using the formula... ,in The offset value is monitored in real time. The predicted value is calculated based on historical trend data. The predicted value can be obtained by a simple linear regression model. The deviation value is used to further analyze the mutation trend. By statistically analyzing the magnitude and direction of the mutation trend and combining the results of the offset rate calculation, the offset rate and prediction deviation parameters are generated.

[0169] S512: Based on the offset rate and prediction deviation parameters, identify the change amplitude of signal mutation trajectory points, combine the temporal continuity of mutation trajectory points and their spatial distribution distance with neighboring trajectory points to determine whether the trajectory points belong to abnormal areas, and generate trajectory point abnormal distribution area parameters by screening and labeling high deviation trajectory points.

[0170] For the deviation value of each abrupt change trajectory point, its anomalous region is determined by considering both temporal continuity and spatial distribution characteristics. Temporal continuity is determined by whether the time interval between adjacent trajectory points exceeds a set range, such as a consecutive point within 10 seconds. The Euclidean distance between adjacent trajectory points is calculated by analyzing their spatial distribution. ,in and The coordinates of adjacent points are used to determine whether an area is an abnormal distribution area by setting a spatial distribution distance threshold. Points with an Euclidean distance greater than the set threshold (20 meters) are removed. The remaining points are classified and labeled based on their deviation values. High deviation trajectory points are further aggregated and statistically analyzed to label the boundary information of the abnormal area. The convex hull algorithm is used to identify the abnormal area surrounded by the trajectory points and generate the boundary information of the abnormal area as the labeling result, thus generating the trajectory point abnormal distribution area parameters.

[0171] S513: Based on the parameters of the abnormal distribution area of ​​trajectory points, construct the labeling rules for stage-specific abnormal trajectories, using the formula:

[0172] ;

[0173] Label abnormal trajectories during the stage and generate the labeling results for abnormal trajectories during the stage.

[0174] in, The abnormal trajectory annotation value represents the stage. This represents the abnormal deviation value of the i-th trajectory point. The weighting factor represents the trajectory point. Represents the coordinates of the trajectory points. The center coordinates of the region representing the abnormal distribution of trajectory points. This represents the total number of trajectory points.

[0175] formula:

[0176] ;

[0177] The advantage of the formula lies in its ability to introduce outlier values. Weighting factors Furthermore, by using the reciprocal of the square root of the spatial distance between trajectory points, the degree of anomaly of trajectory points is correlated with the spatial distance. This allows for a more accurate reflection of the deviation characteristics of spatial distribution when anomaly trajectories are annotated, which is helpful for the accurate annotation and quantitative analysis of anomaly areas.

[0178] Detailed explanation of the formula and its calculation derivation:

[0179] 1. The abnormal deviation value represents the i-th trajectory point. The value range is calculated according to the deviation calculation formula. For example, the deviation value of the first point is 10, the deviation value of the second point is 15, and the deviation value of the third point is 8.

[0180] 2. The weighting factors representing the trajectory points are assumed to be related to the importance of the trajectory points, for example, calculated by normalizing the deviation values, with weights of 0.5, 0.8, and 0.3 respectively.

[0181] 3. Let these be the coordinates of the trajectory points, assuming the coordinates of the trajectory points are (1,2), (3,4), and (5,6). Let (4,4) be the center coordinate of the abnormal distribution area.

[0182] 4. Calculate the distance from each trajectory point to the center point:

[0183] ;

[0184] ;

[0185] ;

[0186] 5. Substitute the parameters into the formula:

[0187] ;

[0188] ;

[0189] ;

[0190] The results show that the labeled value of 14.46 indicates the comprehensive degree of abnormality of the stage abnormal trajectory. The larger the value, the more concentrated the abnormal deviation of the trajectory points and the closer the distribution range. By comparing with the preset benchmark value (20), if it is lower than the benchmark value, the degree of abnormality is considered acceptable. Otherwise, further analysis is required. The labeled abnormal trajectory points can be used for the accurate analysis and early warning of subsequent abnormal areas.

[0191] A physiological data analysis system for post-orthopedic patients, the system being used to execute the aforementioned physiological data analysis method for post-orthopedic patients, the system comprising:

[0192] The data dynamic monitoring module obtains the range of data changes within the sliding window and the frequency changes of adjacent windows based on heart rate fluctuations, changes in blood oxygen saturation, and limb movement amplitude data. It also statistically analyzes the characteristic changes of continuous data segments, adjusts the time span of the sliding window for segmented calculation, and generates segmented physiological dynamic index values.

[0193] The physiological signal clustering analysis module extracts the local density values ​​of heart rate fluctuation and blood oxygen saturation changes based on segmented physiological dynamic index values, calculates the number of high-density points around multiple signal points, determines the distribution distance of high-density points and removes abnormal noise points, summarizes the dynamic change characteristics of high-density areas, and generates time-series clustering feature patterns.

[0194] The recovery phase feature extraction module is based on time-series clustering feature patterns. It calculates the maximum amplitude and frequency interval of the peak signal of heart rate fluctuations in the acute phase, statistically analyzes the slope and fluctuation range of the trend change during the functional reconstruction period, obtains the convergence value and periodic change amplitude features of the signal in the stable period, and generates a set of feature parameters for the recovery phase.

[0195] The phase transition probability analysis module analyzes the correlation changes of characteristic values ​​from the acute phase to the functional reconstruction phase based on the characteristic parameter set of the recovery phase, calculates the key characteristic value transition sequence from the functional reconstruction phase to the stable phase, sets the phase node time correlation characteristics and transition relationship matrix, and generates the recovery phase characteristic transition probability matrix.

[0196] The abnormal trajectory recognition module extracts heart rate fluctuation values ​​and blood oxygen saturation change offset rates based on the feature transition probability matrix of the recovery phase, calculates prediction deviation values ​​and identifies trajectory points of signal mutations, determines whether trajectory points belong to abnormal areas and marks abnormal distribution areas, and generates stage abnormal trajectory labeling results.

[0197] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. An orthopedic post-operative patient physiological data analysis method, comprising: The method comprises the following steps: S1: Collecting the heart rate, blood oxygen saturation and limb activity sensor data of a patient, extracting the heart rate fluctuation value, blood oxygen saturation change value and limb activity amplitude respectively, constructing a multi-dimensional physiological state feature sequence, for each dimension feature in the multi-dimensional physiological state feature sequence, calculating the data change range in the sliding window and the frequency change value of the adjacent window, counting the feature change amplitude of the continuous data segment, dynamically adjusting the window time span to calculate the index in segments, and generating segmented physiological dynamic index values; S2: Based on the segmented physiological dynamic index values, extracting the local density value of the heart rate fluctuation value and the blood oxygen saturation change in the segment, calculating the number of high-density points in the range around the multi-signal point, removing abnormal noise points according to the distribution distance, summarizing the dynamic change characteristics of the high-density area, and generating a time series clustering feature mode; S3: Based on the time series clustering feature mode, calculating the maximum amplitude and frequency interval of the peak signal of the acute phase heart rate fluctuation value, counting the trend change slope and fluctuation range value of the functional reconstruction period, analyzing the signal convergence value and periodic change amplitude characteristics of the stable period, and generating a recovery stage feature parameter set; S4: Based on the recovery stage feature parameter set, analyzing the characteristic value correlation change from the acute phase to the functional reconstruction period, calculating the key characteristic value transfer sequence from the functional reconstruction period to the stable period, setting the time correlation characteristics and transfer relationship matrix of the stage node, and generating a recovery stage feature transfer probability matrix; S5: Based on the recovery stage feature transfer probability matrix, extracting the offset rate and prediction deviation value of the heart rate fluctuation value and the blood oxygen saturation change in the current time period, identifying the signal mutation trajectory point and judging whether it belongs to an abnormal area, labeling the abnormal distribution area of the trajectory point, and generating a stage abnormal trajectory labeling result.

2. The orthopedic post-operative patient physiological data analysis method of claim 1, wherein, The segmented physiological dynamic index values include the data change range in the sliding window, the frequency change value of the adjacent window, and the feature change amplitude of the continuous data segment. The time series clustering feature mode includes the local density value, the number of high-density points, and the dynamic change characteristics. The recovery stage feature parameter set includes the maximum amplitude and frequency interval of the peak signal of the heart rate fluctuation value, the trend change slope and fluctuation range value of the functional reconstruction period, and the signal convergence value and periodic change amplitude characteristics of the stable period. The recovery stage feature transfer probability matrix includes the characteristic value correlation change from the acute phase to the functional reconstruction period, the key characteristic value transfer sequence from the functional reconstruction period to the stable period, and the time correlation characteristics and transfer relationship matrix of the stage node. The stage abnormal trajectory labeling result includes the offset rate and prediction deviation value of the heart rate fluctuation value and the blood oxygen saturation change, the signal mutation trajectory point, and the abnormal distribution area.

3. The orthopedic post-operative patient physiological data analysis method of claim 2, wherein, The acquisition step of the segmented physiological dynamic index values is specifically: S111: Collect the heart rate, blood oxygen saturation and limb activity sensor data of the patient, extract the heart rate fluctuation value, blood oxygen saturation change value and limb activity amplitude respectively, construct a multi-dimensional physiological state feature sequence, for each dimension feature in the multi-dimensional physiological state feature sequence, segment the data according to the time span in the sliding window, calculate the maximum and minimum difference of the heart rate fluctuation range, blood oxygen saturation change range and limb activity amplitude in each sliding window, and obtain the basic physiological parameter change range; S112: Combine the time span change trend of the sliding window, and perform difference operation on the basic physiological parameter change range of the continuous sliding window, calculate the parameter change amplitude of adjacent sliding windows item by item for the heart rate fluctuation range, blood oxygen change range and limb activity amplitude, compare the parameter change amplitude after difference with the time interval of adjacent window frequency, and dynamically select and aggregate according to the gradient amplitude of the fluctuation range to generate a sliding window frequency change feature parameter; S113: Based on the sliding window frequency change feature parameter, the time span of the sliding window is dynamically optimized and adjusted, the feature value of the segmented data is recalculated, and the segmented physiological dynamic index value is generated by using the formula: ； The acquisition step of the time sequence clustering feature mode is specifically: wherein, representing a segmented physiological dynamic indicator value, representing a weight coefficient of a sliding window, representing a current base physiological parameter variation range, representing a previous window variation range, representing a weight coefficient of a time span, representing a current window time span, representing a previous window time span, representing a number of sliding windows, representing a number of time spans.

4. The orthopedic post-operative patient physiological data analysis method of claim 3, wherein, S211: Based on the segmented physiological dynamic index value, extract the local density value of the heart rate fluctuation value and the blood oxygen saturation change, calculate the number of high-density points in a fixed range for each signal point, and generate a local high-density point distribution parameter; S212: Based on the local high-density point distribution parameter, calculate the density change rate between multiple signal points according to the distribution distance between the signal points, remove the signal points with a distribution distance higher than a threshold value, and perform difference on the remaining point density value and its dynamic gradient to obtain a high-density region dynamic change parameter; S213: Based on the high-density region dynamic change parameter, the time sequence dynamic characteristics of signal point distribution are induced to generate a time sequence clustering feature mode by using the formula: The acquisition step of the recovery stage feature parameter set is specifically: ； S311: Based on the time sequence clustering feature mode, extract the maximum amplitude and frequency interval of the acute phase heart rate fluctuation value peak signal, filter high-amplitude signal data, calculate the amplitude distribution range and frequency interval difference of the acute phase peak signal, and generate an acute phase peak feature parameter; wherein, representing a time-series clustering feature pattern, representing a local density value of the i-th point in the high-density region, representing a horizontal spatial coordinate of the i-th point, representing a vertical spatial coordinate of the i-th point, representing a horizontal spatial coordinate of the center point of the high-density region, representing a vertical spatial coordinate of the center point of the high-density region, is a weight factor of the i-th point, is a dynamic gradient value of the density change of the signal point, is a weight coefficient of the dynamic gradient of the density change, is the number of signal points in the high-density region, is the total number of all screening points, , is an iteration index of the summation symbol, respectively representing the iteration range of the points in the high-density region and the screening points.

5. The orthopedic post-operative patient physiological data analysis method of claim 4, wherein, S312: Based on the acute phase peak feature parameter, combine the time sequence data of the functional reconstruction period signal, calculate the slope of the trend change in the functional reconstruction period, filter high-slope change segments by setting a slope change threshold, segment and difference the fluctuation range value of the high-slope change segment, and combine the fluctuation range and the change slope to generate a functional reconstruction period trend feature parameter; S313: Based on the functional reconstruction period trend feature parameter, combine the stable period signal data, analyze the convergence value and periodic change average amplitude of the stable period signal, and generate a recovery stage feature parameter set by using the formula: The acquisition step of the recovery stage feature transition probability matrix is specifically: ​ ； ​ wherein, representing the set of characteristic parameters of the recovery phase, representing the amplitude of the signal of the stable phase, representing the frequency of the signal, representing the period of the signal, is a weight factor for the amplitude, is the absolute value of the signal convergence value, is the adjustment coefficient of the signal convergence value, representing the number of characteristic signals of the amplitude, representing the total number of convergence signals.

6. The orthopedic post-operative patient physiological data analysis method of claim 5, wherein, ​ S411: Based on the recovery stage characteristic parameter set, the correlation change of the characteristic values from the acute stage to the functional reconstruction stage is analyzed, the dynamic change range of the characteristic values of the acute stage and the transfer trend of the characteristic values of the functional reconstruction stage are extracted, the correlation transfer proportion of the characteristic values of the two stages is calculated, and a stage characteristic value transfer relationship parameter is generated; S412: Based on the stage characteristic value transfer relationship parameter, in combination with the characteristic value sequence from the functional reconstruction stage to the stable stage, the key characteristic value transfer sequence from the functional reconstruction stage to the stable stage is counted, the time correlation characteristics among multiple characteristic values in the transfer sequence are calculated, the characteristic sequences with correlation values lower than a preset threshold are removed, and a time correlation characteristic parameter is generated; S413: Based on the time correlation characteristic parameter, a transfer relationship matrix between stage nodes is constructed, and a formula is used to generate a recovery stage characteristic transfer probability matrix; ； The acquisition step of the stage abnormal trajectory labeling result is specifically: wherein, is the transition probability value in the recovery phase feature transition probability matrix, is the time-dependent feature value of the kth phase, is the weight parameter of the time-dependent feature, is the feature distance between the phase nodes, is the adjustment factor of the feature distance, is the total number of phase nodes.

7. The orthopedic post-operative patient physiological data analysis method of claim 6, wherein, S511: Based on the recovery stage characteristic transfer probability matrix, the offset rate of the heart rate fluctuation value and the blood oxygen saturation change in the current time period is extracted, the prediction deviation value of the signal is calculated through point-by-point comparison of the offset rate with a preset reference range, the signal characteristic change amplitude and the offset trend are counted, and an offset rate and prediction deviation parameter is generated; S512: Based on the offset rate and prediction deviation parameter, the change amplitude of the signal mutation trajectory point is identified, the time continuity of the mutation trajectory point and the spatial distribution distance of the adjacent trajectory point are combined, it is judged whether the trajectory point belongs to an abnormal area, the trajectory point abnormal distribution area parameter is generated through screening and labeling of high deviation trajectory points; S513: Based on the trajectory point abnormal distribution area parameter, a stage abnormal trajectory labeling rule is constructed, and a formula is used to label the stage abnormal trajectory and generate a stage abnormal trajectory labeling result; The orthopedic postoperative patient physiological data analysis method according to any one of claims 1-7, the system comprises: ； The data dynamic monitoring module acquires the data change range in the sliding window and the frequency change value of the adjacent window based on the heart rate fluctuation value, the blood oxygen saturation change and the limb activity amplitude data, counts the characteristic change amplitude of the continuous data segment, adjusts the sliding window time span segmentation calculation, and generates segmented physiological dynamic index values; wherein, representing a phase abnormal trajectory label value, representing an abnormal deviation value of the i-th trajectory point, representing a weight factor of the trajectory point, representing a coordinate of the trajectory point, representing a center coordinate of the trajectory point abnormal distribution region, representing a total number of the trajectory points.

8. An orthopedic post-operative patient physiological data analysis system, comprising: The physiological signal clustering analysis module extracts the local density value of the heart rate fluctuation value and the blood oxygen saturation change based on the segmented physiological dynamic index values, calculates the number of high-density points around multiple signal points, judges the high-density point distribution distance and removes abnormal noise points, induces the dynamic change characteristics of the high-density area, and generates a time sequence clustering feature mode; The recovery stage characteristic extraction module calculates the peak signal maximum amplitude and frequency interval of the acute stage heart rate fluctuation value based on the time sequence clustering feature mode, counts the trend change slope and fluctuation range value of the functional reconstruction stage, acquires the signal convergence value and periodic change amplitude characteristics of the stable stage, and generates a recovery stage characteristic parameter set; ​ ​ The stage transition probability analysis module analyzes the characteristic value correlation change from the acute stage to the functional reconstruction stage based on the recovery stage characteristic parameter set, calculates the key characteristic value transition sequence from the functional reconstruction stage to the stable stage, sets the stage node time correlation characteristics and the transition relationship matrix, and generates the recovery stage characteristic transition probability matrix. The abnormal trajectory identification module extracts the heart rate fluctuation value and the blood oxygen saturation change deviation rate based on the recovery stage characteristic transition probability matrix, calculates the prediction deviation value and identifies the signal mutation trajectory point, judges whether the trajectory point belongs to the abnormal area and labels the abnormal distribution area, and generates the stage abnormal trajectory labeling result.

Images