Medical rehabilitation data assessment method and system based on multi-source data analysis

Abstract

The invention belongs to the technical field of medical rehabilitation data evaluation, and particularly relates to a medical rehabilitation data evaluation method and system based on multi-source data analysis, and the method comprises the steps: obtaining evaluation index data from historical rehabilitation data; performing deviation statistical operation on the evaluation index data, and dividing the evaluation index data into first screening data and second screening data according to whether the evaluation index data meets a preset stability condition; performing validity correction analysis on the second screening data to generate corrected second screening data; performing comprehensive evaluation based on the first screening data and the corrected second screening data to generate a target rehabilitation result covering an evaluation period; according to the method, the data stability is evaluated by executing the deviation statistical operation, and the validity correction analysis is executed on the second screening data to replace direct elimination, so that the evaluation accuracy is improved on the basis of keeping the data integrity.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of medical rehabilitation data evaluation, and particularly relates to a medical rehabilitation data evaluation method and system based on multi-source data analysis. BACKGROUND

[0002] With the deep integration of modern medicine and information technology, data-driven quantitative evaluation has become a key link in guiding the rehabilitation treatment process. In the fields of chronic disease management, nervous system function recovery, and sports injury rehabilitation, scientific analysis of patient rehabilitation data can provide objective basis for formulating and optimizing individualized treatment plans, and plays an important role in improving treatment effect and improving patient quality of life.

[0003] However, in the existing medical rehabilitation data evaluation method, the strategy for processing existing data fails to reasonably balance the processing relationship between stable data and abnormal data. In order to avoid errors, abnormal data is simply excluded, resulting in incomplete evaluation data set, which cannot fully reflect the rehabilitation process, so that the data set used for comprehensive evaluation cannot balance the integrity and accuracy, and further affects the reliability of the target rehabilitation result. At the same time, for the existing abnormal data, there is a lack of scientific classification mechanism and targeted correction means, without type division according to the specific characteristics of abnormal data, only relying on a single and universal processing method, which is difficult to effectively retain the effective information in the original data.

[0004] Therefore, how to propose a new medical rehabilitation data evaluation method and system based on multi-source data analysis is a problem to be solved at present. SUMMARY

[0005] The purpose of the present application is to provide a medical rehabilitation data evaluation method and system based on multi-source data analysis, which can perform volatility evaluation and stability evaluation on the grouping information in the medical rehabilitation data, and further perform verification processing on the abnormal data, thereby completing multi-dimensional analysis of the medical rehabilitation data to improve the effectiveness of the rehabilitation data.

[0006] The present application is realized by the following technical scheme: a medical rehabilitation data evaluation method based on multi-source data analysis, the method comprising: obtaining an evaluation index parameter; According to the evaluation index parameter, the historical rehabilitation data is collected, and the evaluation index data is obtained from the historical rehabilitation data; Performing a deviation statistical operation on the evaluation index data to divide the evaluation index data into first screening data and second screening data according to whether the evaluation index data meets a preset stability condition; Performing an effectiveness correction analysis on the second screening data to generate corrected second screening data; and performing comprehensive evaluation based on the first screening data and the second screening data after correction to generate the target rehabilitation result covering the evaluation period.

[0007] In one embodiment, performing the deviation statistical operation comprises: obtaining baseline data, wherein the baseline data is an arithmetic mean of a maximum value and a minimum value in the evaluation index data; calculating a deviation amount of the evaluation index data from the baseline data; and comparing the deviation amount with a preset zero-tolerance threshold to divide the evaluation index data into the first screening data and the second screening data.

[0008] In one embodiment, performing the effectiveness correction analysis comprises: identifying abnormal data points in the second screening data, and determining a correction scheme according to a type of the abnormal data points, wherein the type of the abnormal data points includes a first abnormal type and a second abnormal type classified based on whether the data points meet a preset extreme value condition; and applying the correction scheme to generate the second screening data after correction.

[0009] In one embodiment, applying the correction scheme comprises: for the first abnormal type and the second abnormal type, respectively selecting a model with the highest matching degree from a preset candidate correction model library as a first correction model and a second correction model; and generating correction data using the first correction model and the second correction model to replace the corresponding abnormal data points in the original second screening data.

[0010] The present application also provides a medical rehabilitation data evaluation system based on multi-source data analysis, comprising: a data acquisition module for collecting historical rehabilitation data and obtaining evaluation index data according to the acquired to-be-evaluated index parameters; a data division module for performing a deviation statistical operation to divide the evaluation index data into first screening data and second screening data according to a preset stability condition; a data correction module for performing an effectiveness correction analysis on the second screening data divided by the data division module to generate the second screening data after correction; and a comprehensive evaluation module for performing comprehensive evaluation based on the first screening data and the second screening data after correction generated by the data correction module to generate the target rehabilitation result.

[0011] In one embodiment, the data division module is configured to: determine baseline data of the evaluation index data, wherein the baseline data is an arithmetic mean of a maximum value and a minimum value in the evaluation index data; computing a deviation amount of the evaluation index data from the baseline data; and comparing the deviation amount with a preset zero-tolerance threshold to complete the division of the first screening data and the second screening data.

[0012] In one embodiment, the data correction module is configured to: identify abnormal data points in the second screening data, and determine a correction scheme according to a type of the abnormal data points, wherein the type of the abnormal data points includes a first abnormal type and a second abnormal type classified based on whether the data points satisfy a preset extreme value condition; and apply the correction scheme to generate the corrected second screening data.

[0013] In one embodiment, when applying the correction scheme, the data correction module is further configured to: select, for the first abnormal type and the second abnormal type, a model with the highest matching degree from a preset candidate correction model library as a first correction model and a second correction model, respectively; and generate correction data using the first correction model and the second correction model to replace the corresponding abnormal data points in the original second screening data.

[0014] Beneficial effects: The present application can perform deviation statistical operation on evaluation index data, dynamically determine the normalization processing interval according to the trend parameters between adjacent data points, and then normalize the data in the interval to obtain the deviation amount of the quantized fluctuation degree, which can objectively quantify the stability of the evaluation index data at the initial stage of data processing, effectively identify the data that does not meet the preset stability condition, and lay a foundation for subsequent analysis. The present application performs effectiveness correction analysis on the second screening data selected by the deviation statistical operation, can divide the abnormal data points into the first abnormal type and the second abnormal type according to the preset extreme value condition, and match the corresponding correction model for each type from the candidate correction model library to generate correction data. Thus, by correcting different types of abnormal data points, the effective information of the original data can be preserved to the greatest extent, and the data integrity and evaluation accuracy are improved.

[0015] When generating the target rehabilitation result, the present application combines the first screening data with the corrected second screening data obtained by the effectiveness correction analysis, and performs comprehensive evaluation based on the combined data set. Through this double-path processing strategy of preserving stable data and correcting abnormal data, it is ensured that the data used for evaluation is neither distorted by abnormal data points nor incomplete due to data rejection, thereby effectively ensuring the data integrity and evaluation accuracy of the target rehabilitation result. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 is a flowchart of embodiment 1 of the present application. DETAILED DESCRIPTION

[0017] In order to make the objectives, technical solutions and advantages of the present application clearer, further detailed description will be given to the present application in combination with specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and should not be used to limit the protection scope of the present application.

[0018] Embodiment 1 Please refer to Figure 1 As shown in the figure, the embodiment provides a medical rehabilitation data evaluation method based on multi-source data analysis. The method performs fluctuation evaluation operation and deviation statistical operation on multiple data groups constructed from historical rehabilitation data, and performs effectiveness correction analysis on the identified abnormal data points, thereby completing multi-dimensional analysis of medical rehabilitation data to improve the effectiveness of rehabilitation data. The method specifically includes the following steps: Step S100: Obtain an evaluation index parameter.

[0019] Specifically, the evaluation index parameter is obtained, and multiple data groups are constructed according to the evaluation index parameter. The evaluation index parameter is used to define the scope of the evaluation task. The evaluation index parameter includes an evaluation index, an evaluation start time and an evaluation end time. The evaluation start time and the evaluation end time are used to define the evaluation period.

[0020] It should be noted that the evaluation period is a time interval for data collection and analysis, which is defined by a specific evaluation start time and a specific evaluation end time. According to the evaluation index parameter, the historical rehabilitation data of one or more subjects within the evaluation period is collected from the historical database. According to the preset rule (for example, according to the subject, according to the rehabilitation therapy type or according to the specific time window within the evaluation period), the historical rehabilitation data is constructed into multiple data groups.

[0021] Step S200: Collect historical rehabilitation data according to the evaluation index parameter, and obtain evaluation index data from the historical rehabilitation data.

[0022] Specifically, a fluctuation evaluation operation is performed on each data set to obtain a fluctuation degree thereof, the fluctuation evaluation operation being applied to a quantification analysis process of each data set for measuring a fluctuation amplitude of the evaluation index varying with time. The fluctuation degree, as a quantification result of the fluctuation evaluation operation, represents the stability or fluctuation of the data within a data set, and for quantifying the fluctuation of data, the fluctuation evaluation operation needs to be performed on the evaluation index data in each data set. Specifically, the evaluation index data within a data set is arranged according to its corresponding time stamp to form a time sequence, and then adjacent data points are extracted from the time sequence in sequence, the numerical difference between the latter data point and the former data point is calculated, and the numerical difference is defined as a trend parameter for representing the value of the change direction and amplitude between two continuous data points in the time sequence.

[0023] Further, to realize the unified fluctuation measurement of indexes with different dimensions and numerical ranges, normalization processing needs to be performed, and the normalization processing interval is dynamically determined according to the trend parameter. The normalization processing interval is a data range dynamically determined for normalization, and the range is established according to the numerical value of the trend parameter to ensure consistent scaling between different dimensions. Specifically, the data point appearing earlier in the time sequence is defined as a reference node, and the data point appearing later is defined as a bias node, wherein the reference node is the data point with an earlier time stamp in a pair of adjacent data points in the time sequence, and the bias node is the data point with a later time stamp in a pair of adjacent data points in the time sequence. When the trend parameter is a non-positive value, the normalization processing interval is constructed with the bias node as the starting point, and when the trend parameter is a positive value, the normalization processing interval is constructed with the reference node as the starting point.

[0024] Further, after determining the normalization processing interval, normalization processing is performed on the evaluation index data in the interval, all evaluation index data in the normalization processing interval is obtained as the data to be processed, the minimum value of the data to be processed is determined, and the minimum value is defined as a normalization base value. The normalization base value is the minimum value of all evaluation index data in the given normalization processing interval, and is used as a reference for normalization calculation. The difference between each data to be processed and the normalization base value is calculated, and each difference is defined as an offset parameter.

[0025] Further, the fluctuation degree of the data set is calculated based on the offset parameters, and a preset normalization reference threshold and a risk reference threshold are determined. The normalization reference threshold is used to screen the offset parameters to identify those values representing a significant deviation from the baseline. The risk reference threshold defines the upper limit of the acceptable level of data fluctuation. Each offset parameter is compared with the normalization reference threshold, and the offset parameter with a value greater than or equal to the normalization reference threshold is determined as an offset value. The offset value is used to represent a significant fluctuation. The offset value is compared with the risk reference threshold. If the offset value is less than the risk reference threshold, the fluctuation degree of the data set is calculated and output according to a preset fluctuation calculation formula.

[0026] The fluctuation calculation formula is a calculation model for aggregating all offset values less than the risk reference threshold in the data set into a single fluctuation degree score. The specific formula is as follows: ; wherein, represents the fluctuation degree, which is a comprehensive quantitative index representing the volatility of the entire data set. The larger the value, the more intense the fluctuation. represents the number of effective offset values, which is the total number of elements in the set . represents a single effective offset value, which is the th offset value in the set . represents the normalization reference threshold, which is a preset baseline value used to standardize the offset value to measure its relative importance. represents the sensitivity coefficient, which is a preset constant (usually ) used to adjust the sensitivity of the fluctuation degree calculation to large amplitude offset values. If the offset value is greater than or equal to the risk reference threshold, the value of the risk reference threshold is output as the fluctuation degree of the data set. The formula input can be a set of all offset values less than the risk reference threshold for a data set.

[0027] Step S300: Perform a deviation statistical operation on the evaluation index data to divide the evaluation index data into first screening data and second screening data according to whether the evaluation index data meets a preset stability condition. The deviation statistical operation is used to evaluate the deviation of each data point from the central tendency (baseline data) of the data set it belongs to, and accordingly divide the data into stable and unstable categories.

[0028] ​Specifically, baseline data is obtained for each data group, and all evaluation indicator data in the same data group is sorted according to the corresponding time stamp. The maximum value and the minimum value of the sorted evaluation indicator data are determined, and the arithmetic mean of the maximum value and the minimum value is calculated and defined as the baseline data of the data group, which is used to represent a single numerical value of the central tendency of the data group.

[0029] The numerical difference between each data point in the evaluation indicator data and the baseline data is calculated, and the absolute value of the numerical difference is defined as the deviation amount. The deviation amount is compared with a preset zero tolerance threshold, which defines the maximum deviation amount allowed when the data point is determined to be stable. If the deviation amount is less than the zero tolerance threshold, the data point is determined to satisfy the preset stability condition and is divided into first screening data. The first screening data is a subset of the original data points that are determined to be stable, wherein the deviation of each data point from the baseline data of the data group is lower than the preset zero tolerance threshold. If the deviation amount is greater than or equal to the zero tolerance threshold, the data point is determined not to satisfy the preset stability condition and is divided into second screening data, which is a subset of the original data points that are determined to be unstable or possibly abnormal.

[0030] Step S400: performing validity correction analysis on the second screening data to generate corrected second screening data.

[0031] Specifically, the validity correction analysis is used to identify, classify and correct or replace abnormal data points through a model-based strategy. The validity correction analysis first classifies abnormal data points to determine whether each data point in the second screening data satisfies a preset extreme value condition. The preset extreme value condition is used to determine whether a data point is an extreme value compared to its local data neighborhood or overall data distribution, for example, the numerical value of the data point exceeds the range of three standard deviations from the local mean. If the data point satisfies the preset extreme value condition, the data point is classified as a first abnormal type. The first abnormal type is a category of abnormal data points that satisfy the preset extreme value condition, which is usually represented by a significant and sharp peak or valley in the data. If the data point does not satisfy the preset extreme value condition, the data point is classified as a second abnormal type. The second abnormal type is a category of abnormal data points that are classified into the second screening data although they do not satisfy the preset extreme value condition, which is usually represented by a more moderate or sustained deviation.

[0032] Different correction schemes are applied to different types of abnormal data points, and a candidate correction model library is pre-set. The candidate correction model library is a pre-established repository of various mathematical or statistical models (such as interpolation models, regression models, and time series prediction models). These models can be used to generate corrected data values. Data points of the first and second abnormal types are matched to multiple models in the candidate correction model library. The degree of fit between the neighboring data around each data point and the prediction results of each candidate correction model is calculated for each abnormal data point.

[0033] The matching degree is a computational model used to quantitatively evaluate the applicability of a candidate correction model to a specific abnormal data point. This is achieved by measuring the degree of agreement between the model's predictions and the reliable data surrounding the abnormal point. The specific formula is as follows: ; where the input can be a time point Abnormal data points , one that does not contain The set of its own neighboring data points ; A candidate model from the candidate revision model library , the output can be a model For outliers Match score ; Indicates the matching degree, which is a quantitative indicator representing the degree of fit between the candidate model prediction value and the actual adjacent data point. The higher the score, the more suitable the model is for correcting the outlier. Indicates the value of the adjacent data point, which means the outlier The value of a reliable data point in the neighborhood; Indicates the timestamp of the adjacent data points, which means the data points The corresponding timestamp; Represents the model prediction value, which means the candidate model At the time point Output predicted value.

[0034] Further, for the first abnormal type, the model with the highest matching degree is selected from the candidate correction model library as the first correction model, which is a specific model with the highest matching degree for correcting the first abnormal type data points, and for the second abnormal type, the model with the highest matching degree is selected from the candidate correction model library as the second correction model, which is a specific model with the highest matching degree for correcting the second abnormal type data points. The first correction data is generated by using the first correction model, and the second correction data is generated by using the second correction model, and the first correction data and the second correction data are used to replace the data points of the first abnormal type and the second abnormal type at the corresponding time position in the original second screening data to generate the corrected second screening data.

[0035] Step S500: comprehensive evaluation based on the first screening data and the corrected second screening data to generate the target rehabilitation result covering the evaluation period.

[0036] Specifically, the comprehensive evaluation is performed based on the screened and corrected data, the first screening data divided in the above steps is combined with the corrected second screening data to form an evaluation data set, which is a complete data set for the final comprehensive evaluation, formed by combining the stable first screening data and the corrected second screening data. Based on the evaluation data set, the system performs comprehensive evaluation on the subject and generates the target rehabilitation result. The target rehabilitation result is a comprehensive evaluation report or conclusion about the rehabilitation state of the subject in the entire evaluation period generated based on the evaluation data set, and the target rehabilitation result covers the entire evaluation period from the evaluation start time to the evaluation end time.

[0037] Through the description of the above embodiments, the present application can perform complete and systematic analysis and processing on the data groups in the medical rehabilitation data, and introduces an adaptive correction strategy based on model matching in the data correction stage, thereby ensuring the scientificity and accuracy of the overall medical rehabilitation data evaluation.

[0038] Embodiment two The present embodiment provides a medical rehabilitation data evaluation system based on multi-source data analysis, which is used to execute the medical rehabilitation data evaluation method described above. The system can effectively process the fluctuations and abnormal data points that may occur in the data collection process through screening, correction and comprehensive evaluation of historical rehabilitation data, thereby generating more objective and accurate target rehabilitation results to provide reliable data support for clinical diagnosis and rehabilitation plan adjustment. The system can be divided into the following several modules that work cooperatively: The data acquisition module is configured to receive a specified evaluation index parameter in response to an evaluation request triggered by an external or internal system, such as a knee flexion angle or a peak grip strength of a specific subject at a certain rehabilitation stage, collect relevant historical rehabilitation data from a historical database or a data stream according to the evaluation index parameter, and extract time series data or discrete measurement data directly corresponding to the evaluation index from the collected data as evaluation index data for subsequent processing.

[0039] The data division module is configured to receive the evaluation index data provided by the data acquisition module and perform a deviation statistical operation thereon to preliminarily classify the data points according to their stability. In a specific execution process, baseline data is obtained, which is an arithmetic mean value calculated from the maximum and minimum values of the evaluation index data and is used to represent the central tendency of the data set. The deviation of each data point in the evaluation index data from the baseline data is calculated one by one, and the calculated deviation is compared with a preset zero-tolerance threshold. If the deviation of a data point is not greater than the threshold, it indicates that the fluctuation of the data point is within an acceptable range and meets the preset stability condition, and the data point is classified as first screening data. Otherwise, if the deviation is greater than the threshold, it indicates that the data point may have abnormal fluctuations and does not meet the preset stability condition, and the data point is classified as second screening data.

[0040] The data correction module is configured to process the second screening data that may have problems screened out by the data division module, perform validity correction analysis thereon to restore the authenticity of the data, and identify abnormal data points in the second screening data. The identification process is based on whether the data point meets a preset extreme value condition, for example, a data point value is much higher or lower than the average value of its adjacent data points. According to the direction and degree of deviation, the module classifies abnormal data points into different types, such as a first abnormal type (e.g., instantaneous sharp pulse) and a second abnormal type (e.g., data drop or signal loss).

[0041] After determining the type of abnormal data points, a corresponding correction scheme is determined for each type. To apply the correction scheme, a preset candidate correction model library needs to be accessed, which stores multiple data correction algorithms or models, such as linear interpolation, polynomial fitting, and moving average filtering. For the identified first abnormal type and second abnormal type, the highest matching model can be selected from the library as the first correction model and the second correction model, respectively. The evaluation of matching degree can be based on local features and change trends of the data. The selected first correction model and second correction model are used to generate new correction data for the corresponding abnormal data points, and the correction data is used to replace the abnormal data points in the original second screening data, thereby generating corrected second screening data.

[0042] The comprehensive evaluation module is used to receive the first screening data (i.e., the original stable data) from the data division module and the corrected second screening data from the data correction module, merge these two parts of data to form a cleaned and corrected evaluation data set covering the entire evaluation cycle, and perform a comprehensive evaluation based on this evaluation data set. The evaluation method may include calculating key statistical indicators (such as average value, improvement rate, stability coefficient) and generating trend analysis charts, or inputting the data into a more advanced rehabilitation effect prediction model, so as to output a comprehensive target rehabilitation result, which can provide medical staff or subjects with clear and accurate rehabilitation progress feedback.

[0043] This embodiment can automatically complete the in-depth processing and analysis of medical rehabilitation data through the collaborative work of the above-mentioned data acquisition module, data division module, data correction module and comprehensive evaluation module. Through the intelligent data screening and correction mechanism, the accuracy and reliability of the evaluation results are improved. It is suitable for various rehabilitation treatment scenarios that require long-term monitoring and precise evaluation.

[0044] The above are only preferred embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and modifications. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

Claims

1. A medical rehabilitation data evaluation method based on multi-source data analysis, characterized in that: Methods include: Get the parameters of the indicator to be evaluated; Collect historical rehabilitation data according to the parameters of the indicators to be evaluated, and obtain evaluation indicator data from the historical rehabilitation data; performing a deviation statistical operation on the evaluation index data to divide the evaluation index data into first screening data and second screening data according to whether the evaluation index data meets a preset stability condition; performing validity correction analysis on the second screening data to generate corrected second screening data; and performing a comprehensive assessment based on the first screening data and the revised second screening data to generate a target rehabilitation result covering the assessment period.

2. A medical rehabilitation data evaluation method based on multi-source data analysis according to claim 1, characterized in that: The operations for performing deviation statistics include: Obtain baseline data, where the baseline data is the arithmetic mean of the maximum and minimum values ​​in the evaluation indicator data; Calculate the deviation between the evaluation index data and the baseline data; And, the deviation amount is compared with a preset zero tolerance threshold to divide the evaluation index data into first screening data and second screening data.

3. The medical rehabilitation data evaluation method based on multi-source data analysis according to claim 1, characterized in that: Performing a validity revision analysis involves: Identifying abnormal data points in the second screened data and determining a correction solution based on the type of the abnormal data points, wherein the type of the abnormal data points includes a first abnormal type and a second abnormal type classified based on whether the data points meet a preset extreme value condition; and applying the correction scheme to generate corrected second screening data.

4. The medical rehabilitation data evaluation method based on multi-source data analysis according to claim 3, characterized in that: Application fixes include: For the first abnormality type and the second abnormality type, respectively select the model with the highest matching degree from the preset candidate correction model library as the first correction model and the second correction model; Furthermore, the first correction model and the second correction model are used to generate correction data to replace the corresponding abnormal data points in the original second screening data.

5. A medical rehabilitation data evaluation system based on multi-source data analysis, characterized in that: include: A data acquisition module is used to collect historical rehabilitation data and obtain evaluation indicator data based on the acquired evaluation indicator parameters; A data partitioning module is used to perform a deviation statistical operation to divide the evaluation index data into first screening data and second screening data according to a preset stability condition; a data correction module, configured to perform validity correction analysis on the second screening data divided by the data division module to generate corrected second screening data; and a comprehensive evaluation module for performing a comprehensive evaluation based on the first screening data and the corrected second screening data generated by the data correction module to generate a target rehabilitation result.

6. The medical rehabilitation data evaluation system based on multi-source data analysis according to claim 5, characterized in that: The data partitioning module is configured as follows: Determine the baseline data of the evaluation indicator data, wherein the baseline data is the arithmetic mean of the maximum value and the minimum value in the evaluation indicator data; Calculate the deviation between the evaluation index data and the baseline data; And, the deviation amount is compared with a preset zero tolerance threshold to complete the division of the first filtered data and the second filtered data.

7. The medical rehabilitation data evaluation system based on multi-source data analysis according to claim 5, characterized in that: The data correction module is configured as follows: Identifying abnormal data points in the second screened data and determining a correction solution based on the type of the abnormal data points, wherein the type of the abnormal data points includes a first abnormal type and a second abnormal type classified based on whether the data points meet a preset extreme value condition; and applying the correction scheme to generate corrected second screening data.

8. The medical rehabilitation data evaluation system based on multi-source data analysis according to claim 7, characterized in that: When applying the correction scheme, the data correction module is further configured to: For the first abnormality type and the second abnormality type, respectively select the model with the highest matching degree from the preset candidate correction model library as the first correction model and the second correction model; Furthermore, the first correction model and the second correction model are used to generate correction data to replace the corresponding abnormal data points in the original second screening data.

Images