Multi-parameter monitoring data analysis method

Abstract

The invention discloses a multi-parameter monitoring data analysis method, and particularly relates to the technical field of medical monitoring, and the method comprises the following steps: obtaining a plurality of preset physiological parameters, carrying out the classification analysis, and dividing the parameters into a correlation physiological parameter set and an independence physiological parameter set; carrying out joint credibility analysis on the relevance physiological parameter set, identifying reliability hidden dangers, and obtaining a consistency quality index through overall trend consistency analysis; independent credibility analysis is carried out on the independent physiological parameters, and an accurate quality index is obtained through accuracy analysis; based on the quality indexes, the physiological parameter groups are divided into high-quality data and low-quality data, and the low-quality data are recollected to improve the data quality; through the universality and flexibility of the method, the method can adapt to different medical scenes, stable monitoring data quality guarantee is provided, clinical risks caused by data quality problems are effectively reduced, and monitoring data processing is more scientific and efficient.

Description

Technical Field

[0001] The present invention relates to the field of medical monitoring technology, and more specifically, to a multi-parameter monitoring data analysis method. Background Art

[0002] With the advancement of medical technology, physiological parameter monitoring devices have been widely used in clinical applications. These devices can monitor various physiological parameters of patients (such as heart rate, blood pressure, blood oxygen saturation, etc.) in real time, providing important data support for clinical diagnosis and treatment. However, in the actual monitoring process, due to factors such as sensor failure, environmental interference, and equipment errors, the quality and reliability of monitoring data may be insufficient. This situation may lead to incorrect clinical decisions and affect the treatment effect and safety of patients.

[0003] In the prior art, abnormalities in monitoring data are usually identified and processed through simple threshold judgment or manual intervention. However, these methods often have limitations when processing complex monitoring data, and cannot effectively identify potential anomalies in multi-parameter correlation data, and it is difficult to systematically evaluate the reliability of independent physiological parameters. Therefore, it is of great significance to propose a more systematic and accurate monitoring data quality assessment and improvement method. Summary of the invention

[0004] To achieve the above object, the present invention provides the following technical solutions: A multi-parameter monitoring data analysis method comprises the following steps: Acquire multiple preset physiological parameters, and then perform classification analysis to classify the multiple preset physiological parameters into categories to obtain a related physiological parameter set and an independent physiological parameter set; Conduct joint credibility analysis on the associated physiological parameter sets to identify whether each associated physiological parameter set has hidden dangers of insufficient reliability; conduct independent credibility analysis on the independent physiological parameter sets to identify whether each independent physiological parameter has hidden dangers of insufficient reliability; The overall trend consistency analysis is performed on the associated physiological parameter group with hidden dangers of insufficient reliability to obtain the consistency quality index; the accuracy analysis is performed on the independent physiological parameters with hidden dangers of insufficient reliability to obtain the accuracy quality index; The associated physiological parameter group is divided into a high-quality parameter group and a low-quality parameter group based on the consistency quality index, and the independent physiological parameters are divided into high-quality parameters and low-quality parameters based on the accuracy quality index; The data of low-quality parameter groups and low-quality parameters were collected again.

[0005] In a preferred embodiment, the associated physiological parameter set consists of a plurality of associated physiological parameter groups, and each associated physiological parameter group consists of a plurality of associated physiological parameters.

[0006] In a preferred embodiment, a plurality of preset physiological parameters are obtained, and then classified and analyzed, and the plurality of preset physiological parameters are classified into categories to obtain a set of associated physiological parameters and a set of independent physiological parameters, respectively, which refers to: In a fixed time window, multiple preset physiological parameters are obtained, and the correlation coefficient between each pair of physiological parameters is calculated using the Pearson correlation coefficient. The calculation formula is: ; represents the i-th sample value of parameter x, represents the i-th sample value of parameter y, represents the mean value of parameter x, represents the mean value of parameter y, The Pearson correlation coefficient between parameter x and parameter y is represented, and then according to the correlation matrix of the Pearson correlation coefficient, a clustering algorithm is applied to divide the associated physiological parameters and the independent physiological parameters. The parameters clustered together are classified as associated physiological parameters, and the clusters formed by the associated physiological parameters are associated physiological parameter groups. All the associated physiological parameter groups are aggregated to obtain an associated physiological parameter set, and the independent parameters are classified as independent physiological parameters. All the independent physiological parameters are aggregated to obtain an independent physiological parameter set.

[0007] In a preferred embodiment, the logic of applying the clustering algorithm is: Use the inverse of the correlation matrix as the distance matrix for hierarchical clustering. The inverse value is one minus the correlation coefficient. The difference between the two clusters is taken as the starting point, starting from each parameter as a separate cluster, and gradually merging the clusters with the highest similarity. Here, the similarity is defined as the average correlation within the cluster. The cluster tree is truncated at a preset height to form multiple cluster groups. Each cluster group corresponds to a group of associated physiological parameters. All parameters in each cluster group are classified as associated physiological parameters. After the cluster tree is truncated, the remaining physiological parameters that have not been merged into any cluster group, that is, those parameters that have not been effectively merged with other clusters during the clustering process, can be identified as separate physiological parameters.

[0008] In a preferred embodiment, performing a joint credibility analysis on the associated physiological parameter sets to identify whether each associated physiological parameter set has a hidden danger of insufficient reliability refers to: Select a target physiological parameter in the associated physiological parameter group and establish a multi-parameter regression model. The model formula is: ; represents the predicted target physiological parameter, is the intercept term, represents the other physiological parameters in the associated physiological parameter group, i represents the number of the other physiological parameters, is the regression coefficient, is the error term; During the real-time monitoring process, the target physiological parameter values ​​predicted by the multi-parameter regression model are compared with the actual measured values ​​to obtain error values, which are the absolute values ​​of the differences between the predicted target physiological parameter values ​​and the actual measured values. A joint credibility analysis is then performed, i.e., the root mean square error value of all errors is calculated, and then the root mean square error value is compared with the preset hidden danger threshold. If the root mean square error value is greater than or equal to the preset hidden danger threshold, then the associated physiological parameter group has the hidden danger of insufficient reliability. If the root mean square error value is less than the preset hidden danger threshold, then the associated physiological parameter group does not have the hidden danger of insufficient reliability.

[0009] In a preferred embodiment, performing independent credibility analysis on the independent physiological parameter set to identify whether each independent physiological parameter has a hidden danger of insufficient reliability refers to: In a fixed time window, independent credibility analysis is performed using the Z score. The Z score is obtained by subtracting the historical average value of the physiological parameter from the current value of the physiological parameter and then dividing it by the historical standard deviation of the physiological parameter. All Z scores in the fixed time window are then compared with the set threshold range. If all Z scores fall within the set threshold range, then the independent physiological parameter does not have the risk of insufficient reliability. If all Z scores do not fall within the set threshold range, then the independent physiological parameter has the risk of insufficient reliability.

[0010] In a preferred embodiment, the overall trend consistency analysis is performed on the associated physiological parameter group with the hidden danger of insufficient reliability, and the consistency quality index is obtained, which means: For each pair of parameters Pi(t) and Pj(t) in the associated physiological parameter group, the parameters Pi(t) and Pj(t) correspond to the data values ​​of parameters i and j at time t respectively. The distance matrix Dij(ti, tj) of each pair of parameters is calculated using the Euclidean distance calculation formula, and the cumulative distance matrix is ​​constructed. Each element of the cumulative distance matrix represents the minimum cumulative distance from the starting point to the current point: ; Represents the cumulative distance matrix for each pair of parameters Pi(t) and Pj(t); Then, a total cumulative distance matrix C(ti, tj) is obtained based on the cumulative distance matrix of all parameter pairs: ; n represents the total number of different physiological parameters contained in the associated physiological parameter group. Based on the total cumulative distance matrix C(ti, tj), the optimal path that makes all time series match as a whole is found, and the overall MDTW distance is calculated as the consistency quality index.

[0011] In a preferred embodiment, the accuracy quality index obtained by performing accuracy analysis on independent physiological parameters with potential risks of insufficient reliability refers to: The independent physiological parameters are obtained through different preset data sources, and all data sources are synchronized in time to ensure that the measurement data of each data source are consistent at the same time point. The data after time synchronization is expressed as: ; The data value of the independent physiological parameter at time t corresponding to the data source number i is represented. The mean and standard deviation of the measurement data of all data sources corresponding to each moment are calculated and input into the pre-trained machine learning model to obtain the comprehensive difference value. The largest comprehensive difference value is the accuracy quality index.

[0012] In a preferred embodiment, dividing the associated physiological parameter group into a high-quality parameter group and a low-quality parameter group based on the consistency quality index means: comparing the consistency quality index with a preset quality threshold one; if the consistency quality index is greater than or equal to the preset quality threshold one, the associated physiological parameter group is divided into a high-quality parameter group; otherwise, it is divided into a low-quality parameter group.

[0013] In a preferred embodiment, dividing independent physiological parameters into high-quality parameters and low-quality parameters based on the accuracy quality index refers to: The accuracy quality index is compared with the preset quality threshold 2. If the accuracy quality index is greater than or equal to the preset quality threshold 2, the independent physiological parameter is classified as a low-quality parameter, otherwise it is classified as a high-quality parameter.

[0014] Technical effects and advantages of the present invention: The present invention introduces a consistency quality index and an accuracy quality index to conduct a systematic quality assessment of associated physiological parameters and independent physiological parameters. This method can effectively identify potential risks in the data and ensure that the data ultimately used for clinical monitoring is highly reliable, thereby reducing the risk of clinical decision-making due to data errors or inaccuracies. By recollecting and verifying the data of low-quality parameter groups and low-quality parameters, the present invention can greatly improve the overall quality of monitoring data. After the recollected data undergoes a strict quality check, it is ensured that all adopted data meet the preset quality standards, providing more reliable data support for clinical analysis and diagnosis.

[0015] The present invention provides a systematic method for classifying physiological parameter data into two categories: high quality and low quality. This classification method is based on scientific quality indicators (such as consistency quality index and accuracy quality index), which can accurately distinguish the reliability of data and provide a clear basis for further data processing and clinical decision-making. By identifying and reprocessing low-quality data, the present invention effectively reduces the clinical risks caused by data quality problems. Reliable data support can help medical staff make more accurate diagnosis and treatment decisions, thereby improving the treatment effect and safety of patients. The present invention reduces the reliance on manual intervention through automated data quality assessment and re-collection mechanisms. This not only improves the work efficiency of monitoring operations, but also enhances the intelligence of the overall operation, making monitoring data processing more scientific and efficient.

[0016] The method of the present invention can be effectively applied in general wards, intensive care units, or telemedicine environments. Its versatility and flexibility enable it to adapt to different medical scenarios and provide stable monitoring data quality assurance. Through multi-source data acquisition and quality assessment, the present invention can identify abnormal data caused by sensor failure or environmental interference, thereby timely eliminating the impact of equipment failure during the monitoring process and ensuring data stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to facilitate understanding by those skilled in the art, the present invention is further described below in conjunction with the accompanying drawings; Figure 1 The schematic diagram is a schematic diagram of a multi-parameter monitoring data analysis method in the present invention. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] Reference Figure 1 The following examples are obtained: Embodiment 1: A multi-parameter monitoring data analysis method comprises the following steps: Acquire multiple preset physiological parameters, and then perform classification analysis to classify the multiple preset physiological parameters into categories to obtain a related physiological parameter set and an independent physiological parameter set; Conduct joint credibility analysis on the associated physiological parameter sets to identify whether each associated physiological parameter set has hidden dangers of insufficient reliability; conduct independent credibility analysis on the independent physiological parameter sets to identify whether each independent physiological parameter has hidden dangers of insufficient reliability; The overall trend consistency analysis is performed on the associated physiological parameter group with hidden dangers of insufficient reliability to obtain the consistency quality index; the accuracy analysis is performed on the independent physiological parameters with hidden dangers of insufficient reliability to obtain the accuracy quality index; The associated physiological parameter group is divided into a high-quality parameter group and a low-quality parameter group based on the consistency quality index, and the independent physiological parameters are divided into high-quality parameters and low-quality parameters based on the accuracy quality index; The data of low-quality parameter groups and low-quality parameters were collected again.

[0020] The associated physiological parameter set is composed of multiple associated physiological parameter groups, and each associated physiological parameter group is composed of multiple associated physiological parameters. Obtaining multiple preset physiological parameters, and then performing classification analysis, classifying the multiple preset physiological parameters into categories, and obtaining the associated physiological parameter set and the independent physiological parameter set respectively refers to: In a fixed time window, multiple preset physiological parameters are obtained, and the correlation coefficient between each pair of physiological parameters is calculated using the Pearson correlation coefficient. The calculation formula is: ; represents the i-th sample value of parameter x, represents the i-th sample value of parameter y, represents the mean value of parameter x, represents the mean value of parameter y, The Pearson correlation coefficient between parameter x and parameter y is represented, and then according to the correlation matrix of the Pearson correlation coefficient, a clustering algorithm is applied to divide the associated physiological parameters and the independent physiological parameters. The parameters clustered together are classified as associated physiological parameters, and the clusters formed by the associated physiological parameters are associated physiological parameter groups. All the associated physiological parameter groups are aggregated to obtain an associated physiological parameter set, and the independent parameters are classified as independent physiological parameters. All the independent physiological parameters are aggregated to obtain an independent physiological parameter set.

[0021] The logic of applying the clustering algorithm is: Use the inverse of the correlation matrix as the distance matrix for hierarchical clustering. The inverse value is one minus the correlation coefficient. , starting with each parameter as a separate cluster, gradually merge the clusters with the highest similarity. Here, similarity is defined as the average correlation within the cluster. The tree is truncated at a preset height of the cluster tree to form multiple cluster groups. Each cluster group corresponds to a group of associated physiological parameters. All parameters in each cluster group are classified as associated physiological parameters. After the cluster tree is truncated, the remaining physiological parameters that have not been merged into any cluster group, that is, those parameters that have not been effectively merged with other clusters during the clustering process, can be identified as separate physiological parameters. Specifically: In the initial stage of cluster analysis, each physiological parameter is regarded as an independent cluster. As the clustering algorithm is executed, the clusters with the highest similarity will gradually merge to form larger cluster groups. Here, similarity is usually defined as the average correlation within the cluster. In each merging process, the similarity between two clusters is calculated, and the average correlation between clusters is defined using the following formula: ; Representation Cluster and Cluster The similarity (i.e., average correlation) between the two clusters, and Respectively represent clusters and Cluster The physiological parameters in these two clusters, express and If the similarity (i.e., average correlation) of two clusters exceeds the preset merging threshold, they are merged into a larger cluster.

[0022] As the merging proceeds, the clustering tree is gradually formed. In order to identify the associated physiological parameter groups, it is necessary to truncate the tree at a preset height (i.e., similarity threshold) of the clustering tree. At this time, each cluster group formed represents a group of associated physiological parameters. After the clustering tree is truncated, the remaining physiological parameters that have not been merged into any clustering group, that is, those parameters that have not been effectively merged with other clusters during the clustering process, can be identified as separate physiological parameters. Specifically, the identification of separate parameters includes the following situations: Parameters not merged into any cluster: After the cluster tree is truncated, the parameters that still exist alone, that is, the parameters that fail to form a cluster group with other parameters, are directly identified as individual physiological parameters. These parameters usually have low correlation with other parameters and cannot be effectively combined with other parameters.

[0023] Clusters with low correlation with other clusters: Even if some parameters are merged into a cluster during the clustering process, if the average correlation of this cluster with other clusters is significantly lower than the set threshold, such a cluster can be considered to be composed of individual physiological parameters. In this case, all parameters in this cluster can be identified as individual physiological parameters.

[0024] Isolated cluster groups: In some cases, some cluster groups formed after the cluster tree truncation are very weakly related to other groups or even close to independent. These cluster groups may also contain isolated physiological parameters, especially if these parameters are only weakly related to other parameter groups in the overall physiological analysis.

[0025] Conducting joint credibility analysis on the associated physiological parameter sets to identify whether each associated physiological parameter set has a hidden danger of insufficient reliability refers to: Select a target physiological parameter in the associated physiological parameter group and establish a multi-parameter regression model. The model formula is: ; represents the predicted target physiological parameter, is the intercept term, represents the other physiological parameters in the associated physiological parameter group, i represents the number of the other physiological parameters, is the regression coefficient, is the error term; the purpose of selecting a target physiological parameter from the associated physiological parameter group and establishing a multi-parameter regression model is to use other related physiological parameters to predict the target parameter, so as to evaluate the reliability of the data through the accuracy of the model prediction; During the real-time monitoring process, the target physiological parameter values ​​predicted by the multi-parameter regression model are compared with the actual measured values ​​to obtain error values, which are the absolute values ​​of the differences between the predicted target physiological parameter values ​​and the actual measured values. A joint credibility analysis is then performed, i.e., the root mean square error value of all errors is calculated, and then the root mean square error value is compared with the preset hidden danger threshold. If the root mean square error value is greater than or equal to the preset hidden danger threshold, then the associated physiological parameter group has the hidden danger of insufficient reliability. If the root mean square error value is less than the preset hidden danger threshold, then the associated physiological parameter group does not have the hidden danger of insufficient reliability.

[0026] When selecting a parameter from the group of associated physiological parameters as the target parameter (i.e. the parameter to be predicted) of the multiple linear regression model, the following factors can be considered: First, the target parameter can be selected based on clinical importance. In a group of related physiological parameters, there are usually some parameters that are more clinically critical. For example, in a group of parameters including heart rate, blood pressure, and blood oxygen saturation, blood pressure may be considered a more critical parameter because it directly affects the diagnosis and treatment of many pathophysiological conditions. Therefore, blood pressure can be selected as the target parameter of the regression model.

[0027] Another way to select the target parameter is based on the strength of the correlation between parameters. Parameters that have a high correlation with other parameters can be selected as target parameters first, which can ensure that the regression model can effectively capture the relationship between parameters. If a parameter has a significantly lower correlation with other parameters than other parameters, it can also be considered as a target parameter because it may benefit more from other parameters.

[0028] The target parameter can be selected based on the availability and stability of the data. Some parameters may be easier to obtain stable, continuous data than others. For example, heart rate data is usually easier to collect continuously and has less noise than respiration rate data, so heart rate can be preferred as the target parameter.

[0029] In some cases, preliminary modeling and performance evaluation can be performed on all possible target parameters, and the parameter with the best prediction performance can be selected as the target parameter. This can be done by calculating the model's root mean square error, determination coefficient and other indicators to evaluate the prediction effect of each model, and finally select the target parameter of the model with the best prediction performance.

[0030] Independent physiological parameters refer to physiological parameters that have no obvious correlation with other parameters during monitoring. Due to the independence of these parameters, their reliability cannot be verified by the correlation with other parameters, so a separate analysis method is required. The Z-score method is a statistically based analysis tool used to evaluate whether the changes in independent physiological parameters deviate from the normal range, thereby identifying the reliability risks of the data, and performing independent credibility analysis on the independent physiological parameter set to identify whether each independent physiological parameter has a risk of insufficient reliability. It means: in a fixed time window, independent credibility analysis is performed through the Z-score. The Z-score is obtained by subtracting the historical average value of the physiological parameter from the current value of the physiological parameter, and then dividing it by the historical standard deviation of the physiological parameter. The historical average value is usually calculated based on data over a longer period of time to reflect the normal level of the physiological parameter. The historical standard deviation is used to measure the fluctuation range of the physiological parameter. The Z-score represents the standardized deviation of the current value relative to the historical average value. According to specific clinical or monitoring needs, the threshold range of the Z-score is set. The threshold range is usually set to [-2,2] or [-3,3], depending on the sensitivity requirements of the monitoring. Significance: The Z-score method provides a simple and effective means to evaluate the reliability of independent physiological parameters. By comparing the current data with the historical data in a standardized manner, potential abnormal data can be quickly identified. If the Z score exceeds the set threshold range, it indicates that there may be sensor failure, data noise, or abnormal patient status. This provides an early warning mechanism for clinical monitoring and helps to intervene in a timely manner. Based on the results of the Z score analysis, the monitoring system can be dynamically adjusted, such as adjusting the sensitivity of parameters, checking the status of the device, or performing more frequent data sampling. Through the accumulation and analysis of historical data, the Z score method can help establish a personalized health baseline for patients, help identify abnormal situations more accurately, and avoid false positives and false negatives. Then all Z scores in a fixed time window are compared with the set threshold range. If all Z scores fall within the set threshold range, then the independent physiological parameter does not have the hidden danger of insufficient reliability. If all Z scores do not fall within the set threshold range, then the independent physiological parameter has the hidden danger of insufficient reliability.

[0031] The overall trend consistency analysis of the associated physiological parameter group with the potential risk of insufficient reliability is performed to obtain the consistency quality index, which means: for each pair of parameters Pi(t) and Pj(t) in the associated physiological parameter group, the parameters Pi(t) and Pj(t) correspond to the data values ​​of parameters i and j at time t respectively, and the distance matrix Dij(ti, tj) of each pair of parameters is calculated using the Euclidean distance calculation formula. ; Dij(ti, tj) represents the time point and The Euclidean distance between the parameters Pi and Pj on the , construct a cumulative distance matrix, each element of the cumulative distance matrix represents the minimum cumulative distance from the starting point to the current point: ; Represents the elements in the cumulative distance matrix accumulation for each pair of parameters Pi(t) and Pj(t), representing the distance from the starting point to and The minimum cumulative distance is the minimum path cumulative distance from the starting point to the current point through dynamic time warping (DTW); Indicates the minimum path selection from the previous moment (up, left, diagonal) to the current point, and then based on the cumulative distance matrix of all parameter pairs, a total cumulative distance matrix C(ti, tj) is obtained to measure the overall trend consistency of the entire parameter group. The cumulative distance matrix of all parameter pairs is weighted averaged to obtain: ; in the formula It acts as a normalization factor, rather than a traditional weighted average. This factor is used to normalize the number of parameter pairs, ensuring that all parameter pairs have the same contribution in the summation process. Specifically: In the classic weighted average, each element is usually multiplied by a specific weight factor, which reflects the relative importance of each element in the average. In this formula, This is to ensure that all parameter pairs contribute equally to the total cumulative distance matrix. Since there are n(n−1) / 2 pairs of parameter pairs, this factor is to average their cumulative distances. In fact, this formula is not a weighted average in the strict sense, because different parameter pairs are not assigned different weights, but are treated equally. It is a standardized summation formula that uniformly calculates the cumulative distances of all parameter pairs and then calculates the average value. This formula can be regarded as an equal-weighted average, in which the contribution of each parameter pair is the same, so no explicit weight factor is introduced. n represents the total number of different physiological parameters contained in the associated physiological parameter group. Based on the total cumulative distance matrix C(ti, tj), find the optimal path that makes all time series match as a whole. It is usually solved by dynamic programming and the overall MDTW distance is calculated as the consistency quality index. The Euclidean distance matrix is ​​used to quantify the difference between each pair of physiological parameters at a specific time point and is the basis for further cumulative distance calculation. The cumulative distance matrix captures the dynamic correlation between physiological parameters by considering the changes in different time steps, providing basic data for overall trend analysis. The total cumulative distance matrix is ​​used to evaluate the overall trend consistency of the entire parameter group to avoid excessive influence of a single parameter on the results. MDTW distance, as a consistency quality indicator, is used to evaluate whether the overall change trend of the associated physiological parameter group in the entire time window is in line with expectations, thereby identifying the hidden dangers of insufficient data reliability. The overall MDTW distance can be described as "multivariate dynamic time warping distance" or "multivariate dynamic time warping distance". This indicator is used to evaluate the consistency of the change trends between multiple time series (such as each physiological parameter in the associated physiological parameter group) in the entire time window. Specifically, the MDTW distance dynamically adjusts and matches the data points in different time series to calculate a distance value representing the overall consistency of these time series. The smaller the distance value, the more consistent the change trends of these time series, indicating that the correlation within the parameter group is strong; the larger the distance value, the greater the difference in the change trends of these time series, and there may be data anomalies or inconsistencies. Therefore, the overall MDTW distance can be understood as an indicator to measure the overall consistency of changes in multiple physiological parameters in the time dimension, which is used to evaluate the reliability and data quality of these parameters during the monitoring process.

[0032] The accuracy quality index of independent physiological parameters with potential reliability problems is obtained by performing accuracy analysis. This means that the independent physiological parameters are obtained through different preset data sources such as multiple sensors or devices, and all data sources are synchronized in time to ensure that the measurement data of each data source is consistent at the same time point. The data after time synchronization is expressed as: ; The data value of the independent physiological parameter at the time t corresponding to the data source number i is represented, and the measurement results of each data source are statistically processed to quantify its central tendency and degree of dispersion, and the mean and standard deviation of the measurement data of all data sources corresponding to each moment are calculated, and input into the pre-trained machine learning model. The machine learning model is not specifically limited here. The present invention provides a specific implementation method: the comprehensive difference value is obtained by multiplying the mean value and the standard deviation by the sum of the two product values ​​of the respective preset proportional coefficients, and the preset proportional coefficient is used to adjust the influence of the mean value and the standard deviation on the comprehensive difference value, and finally obtain the comprehensive difference value. The largest comprehensive difference value is the accuracy quality index. When the accuracy quality index is larger, it means that the mean value and / or standard deviation in the comprehensive difference value have reached a higher level at certain time points. This usually indicates that there are large differences between the measurement values ​​of different data sources, and the fluctuation of the measurement data is large or the degree of deviation from the historical average value is high. Therefore, the larger the accuracy quality index, the lower the reliability of the independent physiological parameter may be, that is, there is a large uncertainty or potential problem in the accuracy of the measurement data.

[0033] Dividing the associated physiological parameter group into a high-quality parameter group and a low-quality parameter group based on the consistency quality index means: comparing the consistency quality index with the preset quality threshold one, if the consistency quality index is greater than or equal to the preset quality threshold one, the associated physiological parameter group is divided into a high-quality parameter group, otherwise it is divided into a low-quality parameter group. The consistency quality index is used to measure the overall consistency within the associated physiological parameter group. This index reflects whether the change trends of the parameters in the group are consistent at different time points, and whether these changes meet the expected correlation. By comparing the consistency quality index with the preset quality threshold one, it can be determined whether the measurement data of the parameter group is stable and reliable. High-quality parameter group: When the consistency quality index is less than the preset quality threshold one, it indicates that the change trend within the parameter group is consistent and the data reliability is high. Low-quality parameter group: When the consistency quality index is greater than or equal to the preset quality threshold one, it indicates that there is inconsistency within the parameter group, there may be data anomalies or measurement errors, and the reliability is low.

[0034] Classifying independent physiological parameters into high-quality parameters and low-quality parameters based on the accuracy quality index means: comparing the accuracy quality index with the preset quality threshold 2, if the accuracy quality index is greater than or equal to the preset quality threshold 2, the independent physiological parameter is classified as a low-quality parameter, otherwise it is classified as a high-quality parameter. The accuracy quality index is used to evaluate the measurement accuracy and reliability of independent physiological parameters. The index reflects the consistency of measurement results between different data sources. By comparing the accuracy quality index with the preset quality threshold 2, it can be determined whether the measurement of the independent physiological parameter is reliable. High-quality parameters: When the accuracy quality index is less than the preset quality threshold 2, it indicates that the measurement values ​​of different data sources are consistent and the data accuracy is high. Low-quality parameters: When the accuracy quality index is greater than or equal to the preset quality threshold 2, it indicates that there are large differences between the measurement data, there may be errors or inaccuracies, and the data reliability is low.

[0035] All physiological parameters and low-quality parameter data in the low-quality parameter group are recollected. The main purpose of recollecting data is to correct or supplement the data previously identified as low-quality, so that it can meet reliable quality standards by improving the accuracy and consistency of the data. Only those data that pass the quality inspection after recollection and meet the preset quality threshold will be adopted as the final monitoring data. These data are considered reliable and can provide an accurate and reliable basis for monitoring analysis. For data that still do not meet the quality standards after recollection, further processing or abandonment may be required to avoid erroneous analysis results and decisions. The adopted data will be used for further monitoring data analysis, including abnormality detection and clinical decision support.

[0036] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0037] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0038] Those of ordinary skill in the art will appreciate that the exemplary algorithm steps described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0039] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0040] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A multi-parameter monitoring data analysis method, characterized in that: The following steps are involved: Acquire multiple preset physiological parameters, and then perform classification analysis to classify the multiple preset physiological parameters into categories to obtain a related physiological parameter set and an independent physiological parameter set; Conduct joint credibility analysis on the associated physiological parameter sets to identify whether each associated physiological parameter group has a hidden danger of insufficient reliability; Conduct independent credibility analysis on the independent physiological parameter set to identify whether each independent physiological parameter has the potential risk of insufficient reliability; Perform overall trend consistency analysis on the associated physiological parameter groups with potential reliability problems to obtain consistency quality index; The accuracy of independent physiological parameters with potential reliability problems was analyzed to obtain the accuracy quality index; Based on the consistency quality index, the associated physiological parameter group was divided into high-quality parameter group and low-quality parameter group. Based on the accuracy quality index, the independent physiological parameters were divided into high-quality parameters and low-quality parameters, and the data of the low-quality parameter group and low-quality parameters were re-collected.

2. A multi-parameter monitoring data analysis method according to claim 1, characterized in that: The associated physiological parameter set is composed of a plurality of associated physiological parameter groups, and each associated physiological parameter group is composed of a plurality of associated physiological parameters.

3. A multi-parameter monitoring data analysis method according to claim 2, characterized in that: Obtaining multiple preset physiological parameters, and then performing classification analysis, classifying the multiple preset physiological parameters into categories, and obtaining the associated physiological parameter set and the independent physiological parameter set respectively refers to: In a fixed time window, multiple preset physiological parameters are obtained, and the correlation coefficient between each pair of physiological parameters is calculated using the Pearson correlation coefficient. The calculation formula is: ; represents the i-th sample value of parameter x, represents the i-th sample value of parameter y, represents the mean value of parameter x, represents the mean value of parameter y, The Pearson correlation coefficient between parameter x and parameter y is represented, and then according to the correlation matrix of the Pearson correlation coefficient, a clustering algorithm is applied to divide the associated physiological parameters and the independent physiological parameters. The parameters clustered together are classified as associated physiological parameters, and the clusters formed by the associated physiological parameters are associated physiological parameter groups. All the associated physiological parameter groups are aggregated to obtain an associated physiological parameter set, and the independent parameters are classified as independent physiological parameters. All the independent physiological parameters are aggregated to obtain an independent physiological parameter set.

4. A multi-parameter monitoring data analysis method according to claim 3, characterized in that: The logic of applying the clustering algorithm is: Use the inverse of the correlation matrix as the distance matrix for hierarchical clustering. The inverse value is one minus the correlation coefficient. The difference between the two clusters is taken as the starting point, starting from each parameter as a separate cluster, and gradually merging the clusters with the highest similarity. Here, the similarity is defined as the average correlation within the cluster. The cluster tree is truncated at a preset height to form multiple cluster groups. Each cluster group corresponds to a related physiological parameter group. All parameters in each cluster group are classified as related physiological parameters. After the cluster tree is truncated, the remaining physiological parameters that have not been merged into any cluster group are classified as separate physiological parameters.

5. A multi-parameter monitoring data analysis method according to claim 4, characterized in that: Conducting joint credibility analysis on the associated physiological parameter sets to identify whether each associated physiological parameter set has a hidden danger of insufficient reliability refers to: Select a target physiological parameter in the associated physiological parameter group and establish a multi-parameter regression model. The model formula is: ; represents the predicted target physiological parameter, is the intercept term, represents the other physiological parameters in the associated physiological parameter group, i represents the number of the other physiological parameters, is the regression coefficient, is the error term; During the real-time monitoring process, the target physiological parameter values ​​predicted by the multi-parameter regression model are compared with the actual measured values ​​to obtain error values, which are the absolute values ​​of the differences between the predicted target physiological parameter values ​​and the actual measured values. A joint credibility analysis is then performed, i.e., the root mean square error value of all errors is calculated, and then the root mean square error value is compared with the preset hidden danger threshold. If the root mean square error value is greater than or equal to the preset hidden danger threshold, then the associated physiological parameter group has the hidden danger of insufficient reliability. If the root mean square error value is less than the preset hidden danger threshold, then the associated physiological parameter group does not have the hidden danger of insufficient reliability.

6. A multi-parameter monitoring data analysis method according to claim 5, characterized in that: Conducting independent credibility analysis on independent physiological parameter sets to identify whether each independent physiological parameter has a hidden danger of insufficient reliability refers to: In a fixed time window, independent credibility analysis is performed using the Z score. The Z score is obtained by subtracting the historical average value of the physiological parameter from the current value of the physiological parameter and then dividing it by the historical standard deviation of the physiological parameter. All Z scores in the fixed time window are then compared with the set threshold range. If all Z scores fall within the set threshold range, then the independent physiological parameter does not have the risk of insufficient reliability. If all Z scores do not fall within the set threshold range, then the independent physiological parameter has the risk of insufficient reliability.

7. A multi-parameter monitoring data analysis method according to claim 6, characterized in that: The consistency quality index is obtained by analyzing the overall trend consistency of the associated physiological parameter groups with potential risks of insufficient reliability: For each pair of parameters Pi(t) and Pj(t) in the associated physiological parameter group, the parameters Pi(t) and Pj(t) correspond to the data values ​​of parameters i and j at time t respectively. The distance matrix Dij(ti, tj) of each pair of parameters is calculated using the Euclidean distance calculation formula, and the cumulative distance matrix is ​​constructed. Each element of the cumulative distance matrix represents the minimum cumulative distance from the starting point to the current point: ; Represents the cumulative distance matrix for each pair of parameters Pi(t) and Pj(t); Then, a total cumulative distance matrix C(ti, tj) is obtained based on the cumulative distance matrix of all parameter pairs: ; n represents the total number of different physiological parameters contained in the associated physiological parameter group. Based on the total cumulative distance matrix C(ti, tj), the optimal path that makes all time series match as a whole is found, and the overall MDTW distance is calculated as the consistency quality index.

8. A multi-parameter monitoring data analysis method according to claim 7, characterized in that: The accuracy quality index obtained by analyzing the accuracy of independent physiological parameters with potential reliability problems is: The independent physiological parameters are obtained through different preset data sources, and all data sources are synchronized in time to ensure that the measurement data of each data source are consistent at the same time point. The data after time synchronization is expressed as: ; The data value of the independent physiological parameter at time t corresponding to the data source number i is represented. The mean and standard deviation of the measurement data of all data sources corresponding to each moment are calculated and input into the pre-trained machine learning model to obtain the comprehensive difference value. The largest comprehensive difference value is the accuracy quality index.

9. A multi-parameter monitoring data analysis method according to claim 8, characterized in that: Dividing the associated physiological parameter group into a high-quality parameter group and a low-quality parameter group based on the consistency quality index means: comparing the consistency quality index with a preset quality threshold one; if the consistency quality index is greater than or equal to the preset quality threshold one, the associated physiological parameter group is divided into a high-quality parameter group; otherwise, it is divided into a low-quality parameter group.

10. A multi-parameter monitoring data analysis method according to claim 9, characterized in that: The classification of independent physiological parameters into high-quality parameters and low-quality parameters based on the accuracy quality index refers to: The accuracy quality index is compared with the preset quality threshold 2. If the accuracy quality index is greater than or equal to the preset quality threshold 2, the independent physiological parameter is classified as a low-quality parameter, otherwise it is classified as a high-quality parameter.

Images