A method for evaluating disease rehabilitation

Abstract

This invention relates to the field of rehabilitation evaluation technology, specifically a method for evaluating disease rehabilitation. By collecting time-series data and processing missing values ​​and noise in the collected data, the quality of input data can be improved, reducing the impact of abnormal data on evaluation results. A dynamic rehabilitation assessment model composed of LSTM, DNN, and output layers can effectively extract the time-series and correlation features of dynamic vital signs data, more comprehensively and accurately reflecting the patient's rehabilitation status and improving the accuracy, stability, and practicality of disease rehabilitation evaluation.

Description

Technical Field

[0001] The present invention relates to the technical field of rehabilitation evaluation, and particularly to a method for evaluating disease rehabilitation. Background Art

[0002] With the continuous deepening of the aging of the population, the demand for the rehabilitation of chronic diseases, postoperative rehabilitation, and cardiovascular and cerebrovascular diseases is also increasing, and disease rehabilitation management has become an important part of the medical and health field. The rehabilitation process generally has a long cycle, large individual differences, and continuous state changes. Therefore, how to make a timely, objective, and accurate evaluation of the patient's rehabilitation status directly affects the formulation, modification, and implementation of the subsequent rehabilitation plan.

[0003] In the current technology, the evaluation of the patient's rehabilitation status still mainly relies on medical staff to make judgments at a certain moment by observing the patient's physiological indicators, clinical symptoms, scale scores, or follow-up results. This type of evaluation method can reflect the patient's current recovery status to a certain extent, but generally based on discrete time-point data, more focused on static analysis, and cannot well reflect the dynamic evolution characteristics of multiple physical signs during the rehabilitation process. Therefore, when the patient's rehabilitation status shows progressive changes, short-term fluctuations, or simultaneous linkage changes of multiple indicators, the traditional evaluation method is difficult to reflect these changes in a timely and accurate manner.

[0004] During rehabilitation monitoring, patients continuously generate various dynamic physical sign data, namely blood pressure, heart rate, body temperature, and other physiological indicators. These data have obvious time-series characteristics, and there is also a certain correlation and coupling between various dynamic physical signs. In the existing technology, some evaluation methods only rely on one indicator or a simple feature combination for analysis, and do not have the ability to carry out joint modeling and in-depth correlation mining on multi-dimensional dynamic physical sign data. Therefore, it is difficult to accurately reflect the patient's rehabilitation status, and it is easy to cause one-sidedness and low accuracy of the evaluation results.

[0005] Therefore, the existing technology for rehabilitation evaluation focuses on static evaluation, cannot reflect the continuous dynamic changes of the rehabilitation status, has a low degree of comprehensive application of various dynamic physical sign data, is difficult to fully utilize the connection information between various physical signs, does not handle well the problems of missing and noise in the dynamic physical sign time-series data, thus affecting the reliability of the evaluation results. The existing evaluation models have weak expression ability for complex time-series features and non-linear features, resulting in the accuracy of rehabilitation evaluation still有待提高.

[0006] Therefore, a method for evaluating disease rehabilitation needs to be proposed to solve the existing technical defects, namely the lack of dynamics in rehabilitation evaluation, insufficient utilization of multi-dimensional physical signs, poor data robustness, and low evaluation accuracy. Summary of the Invention

[0007] To address the technical problems existing in the background art described above, the present invention provides a method for evaluating disease rehabilitation.

[0008] In a first aspect, the present invention provides a method for evaluating disease rehabilitation, comprising the following steps: Step 1: Collect time-series data of multiple dynamic vital signs; Step 2: Preprocess the time-series data of the collected multiple dynamic vital signs to obtain the preprocessed time-series data of the dynamic features; Step 3: Construct a dynamic rehabilitation assessment model, which includes an LSTM layer, a DNN layer, and an output layer. Step 4: The rehabilitation dynamic assessment model outputs rehabilitation assessment results based on the time-series data of the preprocessed dynamic features.

[0009] Furthermore, step four includes: capturing the long-term trend and short-term fluctuation dependence of each dynamic vital sign data at different time points through an LSTM layer to obtain the temporal characteristics of each dynamic vital sign data. By using DNN layers to perform deep fusion and nonlinear fitting of the temporal features of each dynamic vital sign data, the correlation features between dynamic feature data can be mined out, and the rehabilitation status can be evaluated. The rehabilitation evaluation results are output through the output layer.

[0010] Furthermore, between steps three and four, the following is also included: training the rehabilitation dynamic assessment model.

[0011] Furthermore, the preprocessing of the time-series data of the collected multiple dynamic vital signs includes: Data missing value processing is performed on the time series data of multiple dynamic features collected; Denoising is performed on the time series data after missing data processing.

[0012] Furthermore, the data missing value processing for the time-series data of the collected multiple dynamic features includes: Locate the missing values ​​in the dimensions to be filled, and determine the length and number of missing time periods in the time series data; Randomly select and remove the same number and duration of data as the original missing data to generate missing data samples similar to the original missing data; Use the generated missing data samples to fill in the missing data; The effectiveness of imputing missing data samples is evaluated by combining the original data to determine the validity of missing values, and if invalid, missing data samples are regenerated.

[0013] Furthermore, the denoising process for the time-series data after missing data processing includes the following steps: Select the point to be denoised as i; Determine the size and position of the search window and the domain window; Calculate the similarity between similarity metric point j and point i to be denoised; Slide the second neighborhood window to the next similarity measurement point j; Iterate through all metric points j within the search window and calculate the value of point i after denoising.

[0014] Secondly, the present invention provides a computer-readable storage medium including a stored program that, when the program is running, controls the electrical equipment where the computer-readable storage medium is located to execute the above-described disease rehabilitation evaluation method.

[0015] The beneficial effects of this invention are as follows: This invention improves the quality of input data and reduces the impact of abnormal data on evaluation results by collecting time-series data, processing missing values ​​and denoising the collected time-series data. It uses LSTM layer, DNN layer and output layer to form a rehabilitation dynamic assessment model, which can better extract the time-series features and correlation features of dynamic vital signs data, more comprehensively and accurately reflect the patient's rehabilitation status, and improve the accuracy, stability and practicality of disease rehabilitation evaluation. Attached Figure Description

[0016] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0017] Figure 1 This is a flowchart of the disease rehabilitation evaluation method of the present invention; Detailed Implementation The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0018] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, each technical and scientific term used in these embodiments has the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0020] In this invention, terms such as "upper," "lower," "left," "right," "front," "back," "vertical," "horizontal," "side," and "bottom" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used only to facilitate the description of the structural relationships of the various components or elements of this invention and do not specifically refer to any component or element in this invention. They should not be construed as limiting the invention.

[0021] In this invention, terms such as "fixed connection," "connected," and "linked" should be interpreted broadly, indicating a fixed connection, an integral connection, or a detachable connection; a direct connection or an indirect connection through an intermediate medium. Those skilled in the art can determine the specific meaning of these terms in this invention based on the specific circumstances, and they should not be construed as limitations on the invention.

[0022] Example 1: This invention provides a rehabilitation evaluation method, comprising the following steps: S1: Collect time-series data of multiple dynamic vital signs; among which, multiple dynamic vital signs include: blood pressure data, heart rate data, and temperature data, etc. S2: Preprocess the time-series data of multiple dynamic vital signs collected to obtain the time-series data of the preprocessed dynamic features; Specifically, the following steps are included: S2-1: Handling missing data values ​​in the time-series data of multiple dynamic features collected, including the following steps: A1: Determine the dimensions relevant to the data to be filled through correlation calculation, and use linear interpolation to apply the correlation to these dimensions. arrive Missing values ​​within the interval are filled in. Taking dynamic vital sign data X and dynamic vital sign data Y as an example, the correlation between the two dimensions is calculated using the following formula: ; in, ; When the correlation between dynamic vital sign data X and dynamic vital sign data Y to be filled is greater than the critical value c, then dynamic vital sign data X is considered to be a related variable of dynamic vital sign data Y.

[0023] A2: The following formula is used to perform linear interpolation on the data points to be filled in the target dimension to obtain the fitted value. ; ; in, and These are the time points and estimated values ​​corresponding to the data points to be determined. and It is the time and actual value corresponding to the first valid record point after the data missing period. and It is the time and actual value corresponding to the most recent valid record point before the data missing period.

[0024] A3: Utilizing the property that a single variable will fluctuate under the influence of other variables, the system fluctuation within the blank time period is calculated by the multiple linear regression method; The process of multiple linear regression is as follows: First of all, in arrive Regression calculations are performed on the filled dimension within the interval, and the relationship between dynamic vital sign data Y and all relevant variables X is determined by the following formula: ; From this, we can obtain , and The corresponding regression values ​​are denoted as , , .

[0025] Secondly, the regression results and Perform linear interpolation to obtain the line. The function value corresponding to the time point is denoted as .

[0026] Finally, the magnitude of the system's volatility at this data level is calculated using the following formula: ; Because the degree of volatility is related to the data level, therefore in Fluctuations of the system at a horizontal level and The relationship is: ; That is, system fluctuations can be expressed as: ; A4: Add fluctuations to the original data level.

[0027] Specifically, the following formula is used: ; .

[0028] S2-2: Denoise the time series data after missing data processing.

[0029] Specifically, the following steps are included: B1: Select two windows of fixed size: the search window and the domain window; The search window is used to limit the range of similar points to be found; the neighborhood window is used to determine the neighborhood size between the point to be denoised and the similar point.

[0030] B2: Calculate the similarity between point i to be denoised and point j to be calculated using a weighting factor; B3: By sliding the domain window within the search window, traverse all points within the search window to find the similarity between all points within the search window and the domain centered on the point i to be denoised.

[0031] B4: When the similarity is less than the preset threshold, perform noise reduction processing.

[0032] S3: Construct a dynamic rehabilitation assessment model, wherein the dynamic rehabilitation assessment model includes: an LSTM layer, a DNN layer, and an output layer.

[0033] A complete LSTM network architecture includes an input layer, a forget gate, an update gate, and an output gate. During the learning process, the LSTM network automatically stores historical information that it deems useful. , represents the memory of the current LSTM cell, This represents the memory of the previous LSTM cell. Let represent the input at time t, where "addition" indicates the addition of information and "multiplication" indicates scaling. The activation functions for the forget gate, update gate, and output gate are all Sigmoid, because information processed by the Sigmoid function becomes either 0 or 1, thus determining which information to forget or remember after passing through that node. Furthermore, since the second derivative of tanh has a long distance to zero, it effectively overcomes the gradient vanishing problem during training. The formula for calculating the update gate at each time step is shown below: ; It updates the weight of the gate. It is the bias vector for updating the gate.

[0034] The forgetting rate in the forgetting gate is calculated as follows: ; in, It is the weight of the Forgotten Gate. This is the bias vector of the forget gate; the cell state update at this moment is as follows: ; in, and These are the weights and bias vectors of the hidden units in the forget gate, respectively.

[0035] Using a similar method, the updated output gate information can be obtained, as shown in the following formula: ; ; in, These are the weights of the output gates. It is the bias vector of the output gate.

[0036] Specifically, assuming the number of time points in the time dimension is n, the number of nodes in the LSTM layer will also be n. To address the vanishing gradient problem during training and improve model training speed, the ReLU activation function is used. In the DNN layer, which is essentially a fully connected layer, the number of hidden layers is set according to the amount of data, and the activation function used is Sigmoid. Finally, the number of nodes in the output layer is set according to the categories; if there are n categories, then n nodes are set. After configuring the network structure, a loss function needs to be defined to calculate gradients and optimize them. The root mean square error (RMSE) method is used as the loss function.

[0037] S4: Train the rehabilitation dynamic assessment model; The parameters are configured, including the number of iterations (epochs), the batch size (BatchSize) fed into the network each time, and the validation data used for validation after each epoch. The model's evaluation function, `metics`, is set to "accuracy," which simply measures the ratio of correctly predicted data to the total number of predictions in the test set. The optimizer during training uses the Adam algorithm; training stops when the accuracy continuously increases and stabilizes, meeting the criteria for model termination.

[0038] S5: The rehabilitation dynamic assessment model outputs rehabilitation assessment results based on time-series data of preprocessed dynamic characteristics.

[0039] Specifically, this includes: using DNN layers to perform deep fusion and nonlinear fitting of the temporal features of each dynamic vital sign data to uncover the correlation features between dynamic feature data and to evaluate the rehabilitation status; The rehabilitation evaluation results are output through the output layer.

[0040] Example 2: The present invention provides a computer-readable storage medium, which includes a stored program that, when the program is running, controls the power equipment where the computer-readable storage medium is located to execute the disease rehabilitation evaluation method described in Example 1.

[0041] The same or similar parts between the various embodiments in this specification can be referred to mutually. In particular, the terminal embodiments are basically similar to the method embodiments, so the description is relatively simple, and the relevant parts can be referred to the description in the method embodiments.

[0042] In the embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or units may be electrical, mechanical, or other forms.

[0043] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0044] Additionally, it should be noted that the flowcharts in the accompanying drawings illustrate methods according to embodiments of this disclosure. In the descriptions corresponding to the flowcharts or block diagrams in the drawings, the operations or steps corresponding to different blocks may occur in a different order than disclosed in the description; sometimes, there is no specific order between different operations or steps. For example, two consecutive operations or steps can actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the function involved. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions.

[0045] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for evaluating disease rehabilitation, characterized in that, Includes the following steps: Step 1: Collect time-series data of multiple dynamic vital signs; Step 2: Preprocess the time-series data of the collected multiple dynamic vital signs to obtain the preprocessed time-series data of the dynamic features; Step 3: Construct a dynamic rehabilitation assessment model, which includes an LSTM layer, a DNN layer, and an output layer. Step 4: The rehabilitation dynamic assessment model outputs rehabilitation assessment results based on the time-series data of the preprocessed dynamic features.

2. The rehabilitation assessment system according to claim 1, characterized in that, Step four includes: capturing the long-term trend and short-term fluctuation dependence of each dynamic vital sign data at different time points through an LSTM layer to obtain the temporal characteristics of each dynamic vital sign data. By using DNN layers to perform deep fusion and nonlinear fitting of the temporal features of each dynamic vital sign data, the correlation features between dynamic feature data can be mined out, and the rehabilitation status can be evaluated. The rehabilitation evaluation results are output through the output layer.

3. The disease rehabilitation evaluation method according to claim 1, characterized in that, Between steps three and four, the following also applies: training the rehabilitation dynamic assessment model.

4. The disease rehabilitation evaluation method according to claim 1, characterized in that, The preprocessing of the time-series data of the collected multiple dynamic vital signs includes: Data missing value processing is performed on the time series data of multiple dynamic features collected; Denoising is performed on the time series data after missing data processing.

5. The disease rehabilitation evaluation method according to claim 4, characterized in that, The process of handling missing data values ​​in the time-series data of multiple dynamic features includes: Locate the missing values ​​in the dimensions to be filled, and determine the length and number of missing time periods in the time series data; Randomly select and remove the same number and duration of data as the original missing data to generate missing data samples similar to the original missing data; Use the generated missing data samples to fill in the missing data; The effectiveness of imputing missing data samples is evaluated by combining the original data to determine the validity of missing values, and if invalid, missing data samples are regenerated.

6. The disease rehabilitation evaluation method according to claim 4, characterized in that, The noise reduction process for the time-series data after missing data processing includes the following steps: Select the point to be denoised as i; Determine the size and position of the search window and the domain window; Calculate the similarity between similarity metric point j and point i to be denoised; Slide the second neighborhood window to the next similarity measurement point j; Iterate through all metric points j within the search window and calculate the value of point i after denoising.

7. A computer-readable storage medium comprising a stored program, characterized in that, When the program is running, it controls the electrical equipment containing the computer-readable storage medium to perform the disease rehabilitation evaluation method according to any one of claims 1 to 6.

Images