Clinical data monitoring method based on model training

Abstract

The invention discloses a clinical data monitoring method based on model training, and the method comprises the following steps: 1, carrying out data collection, namely collecting clinical data, analyzing a data source in the collection process, and selecting the data source; 2, performing data preprocessing to obtain preprocessed data; step 3, selecting a modeling method, performing modeling by using the preprocessed data, and performing modeling related data acquisition in the modeling process to obtain a plurality of pieces of modeling related data; 4, processing the multiple pieces of modeling related data to obtain multiple pieces of modeling evaluation information, and carrying out preferential selection on the multiple pieces of modeling evaluation information to obtain an optimal model; and 5, putting the optimal model into use, and carrying out clinical data monitoring. According to the invention, clinical data can be monitored more intelligently and comprehensively.

Description

Technical Field

[0001] The present invention relates to the field of monitoring methods, and in particular to a clinical data monitoring method based on model training. Background Art

[0002] Clinical data monitoring is an important part of disease research and treatment. It involves the continuous and systematic collection, analysis, and evaluation of patients' clinical information to guide clinical decision-making, evaluate treatment efficacy and prognosis, and promote the development of disease research.

[0003] Clinical data monitoring methods will be used during the clinical data monitoring process.

[0004] The existing clinical data monitoring methods are relatively traditional and have poor overall monitoring effects. Therefore, a clinical data monitoring method based on model training is proposed. Summary of the Invention

[0005] The technical problem to be solved by the present invention is: how to solve the problems of existing clinical data monitoring methods, which are relatively traditional and have poor overall monitoring effects. A clinical data monitoring method based on model training is provided to solve the problems of clinical data monitoring methods.

[0006] The present invention solves the above technical problems through the following technical solutions, which include the following steps:

[0007] Step 1: Data collection, i.e. collecting clinical data, analyzing the data sources during the collection process, and selecting the data sources;

[0008] Step 2: Perform data preprocessing and obtain preprocessed data;

[0009] Step 3: Select a modeling method, use the pre-processed data for modeling, collect modeling-related data during the modeling process, and obtain multiple modeling-related data;

[0010] Step 4: Process multiple modeling-related data to obtain multiple modeling evaluation information, select the best from the multiple modeling evaluation information, and obtain the optimal model;

[0011] Step 5: Put the optimal model into use and conduct clinical data monitoring.

[0012] Furthermore, the specific process of step one is as follows:

[0013] S1: Clarify the data collection objectives, i.e., determine the type of clinical data that needs to be collected;

[0014] S2: Analyze data sources, evaluate different data sources, and obtain source evaluation information;

[0015] S3: Analyze the source assessment information and select the final data source for data collection.

[0016] Furthermore, the process of obtaining the source assessment information is as follows:

[0017] Collect data related to the data source, including information on the number of times the data source is used, data transmission speed, and the number of abnormalities at the data source;

[0018] The number of times the data source is used is marked as F1, the data transmission speed is marked as F2, and the number of times the data source does not have an abnormality is marked as F3;

[0019] Assign a correction value M1 to F1, a correction value M2 to F2, and a correction value M3 to F3;

[0020] M3>M2>M1;

[0021] Through the formula F1*M1+F2*M2+F3*M3=Ff, the source evaluation information Ff of a single data source is obtained.

[0022] Furthermore, the specific process of analyzing the source assessment information and selecting the final data source for data collection is as follows:

[0023] Sort the source evaluation information Ff of a single data source in descending order, then collect the amount of data required for modeling, and select the source corresponding to the source evaluation information Ff as the data source according to the amount of data required for modeling.

[0024] Furthermore, the specific process of performing data preprocessing and obtaining preprocessed data is as follows:

[0025] First, perform data cleaning to remove duplicate data, handle missing values ​​and outliers;

[0026] Use Python’s Pandas library for data cleaning;

[0027] Then perform data standardization and normalization, standardize the data and scale the data to the required standard range, and then normalize the data and scale the data to a fixed range;

[0028] After that, data transformation is performed. After necessary transformation of the data, the preprocessed data is obtained.

[0029] Furthermore, the process of obtaining the modeling-related data is as follows:

[0030] The preprocessed data is imported into multiple modeling tools for real-time modeling. During the modeling process, the processor usage, memory usage, and modeling time are collected. The processor usage, memory usage, and modeling time constitute the modeling-related data.

[0031] Furthermore, the process of processing the plurality of modeling-related data to obtain the plurality of modeling evaluation information is as follows:

[0032] Extract the processor usage, memory usage, storage device usage, and modeling time from the modeling-related data of a single modeling tool;

[0033] Scoring the processor occupancy rate to obtain a first score P1. When the processor occupancy rate is less than a preset value a1, the first score P1 is a preset value f1. When the processor occupancy rate is between a1 and a2, the first score P1 is a preset value f2. When the processor occupancy rate is greater than a2, the first score P1 is a preset value f3.

[0034] a1<a2,f1<f2<f3;

[0035] Score the memory usage to obtain a second score P2. When the memory usage is less than the preset value b1, the second score P2 is the preset value h1. When the memory usage is between the preset values ​​b1 and b2, the second score P2 is the preset value h2. When the memory usage is greater than the preset value b2, the second score P2 is the preset value h3.

[0036] b1<b2,h1<h2<h3;

[0037] Score the modeling time to obtain a third score P3. When the modeling time is less than the preset value c1, the third score P3 is the preset value k1. When the modeling time is between the preset values ​​c1 and c2, the third score P3 is the preset value k2. When the modeling time is greater than the preset value c2, the first score P3 is the preset value k3.

[0038] c1<c2,k3<k2<k1;

[0039] Assign a correction value E1 to the first score P1, a correction value E2 to the second score P2, and a correction value E3 to the third score P3;

[0040] E1+E2+E3=1, E3>E2=E1;

[0041] The modeling evaluation information Pp is obtained through the formula P1*E1+P2*E2+P3*E3=Pp;

[0042] The modeling evaluation information of each modeling tool is calculated in sequence, that is, a plurality of modeling evaluation information is obtained.

[0043] The larger the value of the modeling evaluation information Pp is, the better the modeling tool corresponding to the modeling evaluation information is.

[0044] Furthermore, the optimal model is put into use to monitor clinical data. That is, after the optimal model is put into use, the clinical data collected in real time is imported. When the clinical data collected in real time matches the clinical abnormality status in the optimal model, an alarm message is generated to provide a prompt.

[0045] Compared with the existing technology, the present invention has the following advantages: the clinical data monitoring method based on model training can ensure that the collected clinical data types are comprehensive and relevant by clarifying the data collection goals and analyzing the data sources.

[0046] By evaluating the data sources, including the number of uses, data transmission speed, and number of anomalies, and assigning different correction values, the accuracy and reliability of the data can be improved. By performing real-time modeling in multiple modeling tools and collecting modeling-related data such as processor occupancy, memory occupancy, and modeling duration, a comprehensive understanding of the performance of different modeling tools can be obtained. By processing modeling-related data, obtaining modeling evaluation information, and selecting the best based on the score and correction value, the optimal modeling tool can be objectively selected to obtain the optimal model. After the optimal model is put into use, it can monitor clinical data in real time. When the data matches the clinical abnormality in the model, an alarm message can be generated to prompt, realizing more comprehensive and intelligent clinical data monitoring, making this method more worthy of promotion and use. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is an overall flow chart of the present invention. DETAILED DESCRIPTION

[0048] The following is a detailed description of an embodiment of the present invention. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process. However, the protection scope of the present invention is not limited to the following embodiment.

[0049] like Figure 1 As shown, this embodiment provides a technical solution: a clinical data monitoring method based on model training, comprising the following steps:

[0050] Step 1: Data collection, i.e. collecting clinical data, analyzing the data sources during the collection process, and selecting the data sources;

[0051] Step 2: Perform data preprocessing and obtain preprocessed data;

[0052] Step 3: Select a modeling method, use the pre-processed data for modeling, collect modeling-related data during the modeling process, and obtain multiple modeling-related data;

[0053] Step 4: Process multiple modeling-related data to obtain multiple modeling evaluation information, select the best from the multiple modeling evaluation information, and obtain the optimal model;

[0054] Step 5: Put the optimal model into use and conduct clinical data monitoring.

[0055] The specific process of step one is as follows:

[0056] S1: Determine the types of clinical data that need to be collected, such as medical history, diagnostic information, treatment records, imaging data, etc.

[0057] Clarify the purpose of data collection, such as disease screening, diagnosis, prognosis assessment or efficacy monitoring.

[0058] S2: Analyze data sources, evaluate different data sources, and obtain source evaluation information;

[0059] S3: Analyze the source assessment information and select the final data source for data collection.

[0060] The process of obtaining the source assessment information:

[0061] Collect data related to the data source, including information on the number of times the data source is used, data transmission speed, and the number of abnormalities at the data source;

[0062] The number of times the data source is used is marked as F1, the data transmission speed is marked as F2, and the number of times the data source does not have an abnormality is marked as F3;

[0063] Assign a correction value M1 to F1, a correction value M2 to F2, and a correction value M3 to F3;

[0064] M3>M2>M1;

[0065] The formula F1*M1+F2*M2+F3*M3=Ff is used to obtain the source evaluation information Ff of a single data source;

[0066] The above process comprehensively evaluates data sources by collecting data on three dimensions: usage frequency (F1), data transmission speed (F2), and the number of times no anomalies occurred (F3). These three dimensions reflect the frequency of use, transmission efficiency, and stability of the data source, respectively, providing a more comprehensive understanding of the data source's performance.

[0067] By assigning different correction values ​​(M1, M2, M3) to different dimensions and calculating the weighted sum (Ff), this process achieves a comprehensive assessment of data source performance. This comprehensive approach can more accurately reflect the performance of data sources in real-world applications.

[0068] This process allows for the adjustment of the correction values ​​for different dimensions (M1, M2, and M3) based on actual conditions. This means that in practice, the weights of each dimension can be flexibly adjusted based on the specific needs and importance of the data source, resulting in an evaluation result that is more in line with the actual situation.

[0069] Because the process is flexible, it can be applied to different types of data sources and different application scenarios. Whether it is a hospital information system, a clinical research database or a public health database, it can be evaluated through this process.

[0070] This process is based entirely on objective data from the data source (F1, F2, F3) and preset correction values ​​(M1, M2, M3), eliminating the influence of subjective judgment on the evaluation results. This makes the evaluation results more objective and fair.

[0071] Repeatable Verification: Because the process is data-driven, its accuracy can be verified by repeatedly collecting data and calculating the evaluation results. This repeatability ensures the stability and reliability of the evaluation results.

[0072] This process can be automated through programming, greatly improving the efficiency of the evaluation. In practical applications, a program can be written to automatically collect relevant data from the data source and calculate the evaluation results.

[0073] Because the evaluation process is efficient, it can quickly respond to changes in data sources and the needs of new application scenarios. This makes the process more practical and flexible in actual applications.

[0074] The specific process of analyzing the source assessment information and selecting the final data source for data collection is as follows:

[0075] Sort the source evaluation information Ff of a single data source in descending order, then collect the amount of data required for modeling, and select the source corresponding to the source evaluation information Ff as the data source according to the amount of data required for modeling.

[0076] The specific process of performing data preprocessing and obtaining preprocessed data is as follows:

[0077] Data cleaning: remove duplicate data, process missing values, outliers, etc. to ensure the accuracy and completeness of the data.

[0078] Use Python's Pandas library for data cleaning, such as using the drop_duplicates() function to remove duplicate data and the fillna function to handle missing values.

[0079] Data standardization and normalization: Standardize the data and scale the data to a similar range, such as standardizing the data to a standard normal distribution with a mean of 0 and a variance of 1.

[0080] Normalize the data and scale the data to a fixed range, such as scaling the features to the [0,1] interval.

[0081] Use Python's Scikit-learn library for data standardization and normalization, such as using the StandardScaler class for data standardization and the MinMaxScaler class for data normalization.

[0082] Data transformation: Perform necessary transformations on the data, such as polynomial feature construction, to improve the performance of the model.

[0083] Use Python's Scikit-learn library to transform data, such as using the PolynomialFeatures class. After performing necessary transformations on the data, the preprocessed data is obtained.

[0084] The process of obtaining the modeling-related data is as follows:

[0085] The preprocessed data is imported into multiple modeling tools for real-time modeling. During the modeling process, the processor usage, memory usage, and modeling time are collected. The processor usage, memory usage, and modeling time constitute the modeling-related data.

[0086] The process of processing multiple modeling-related data to obtain multiple modeling evaluation information is as follows:

[0087] Extract the processor usage, memory usage, storage device usage, and modeling time from the modeling-related data of a single modeling tool;

[0088] Scoring the processor occupancy rate to obtain a first score P1. When the processor occupancy rate is less than a preset value a1, the first score P1 is a preset value f1. When the processor occupancy rate is between a1 and a2, the first score P1 is a preset value f2. When the processor occupancy rate is greater than a2, the first score P1 is a preset value f3.

[0089] a1<a2,f1<f2<f3;

[0090] Score the memory usage to obtain a second score P2. When the memory usage is less than the preset value b1, the second score P2 is the preset value h1. When the memory usage is between the preset values ​​b1 and b2, the second score P2 is the preset value h2. When the memory usage is greater than the preset value b2, the second score P2 is the preset value h3.

[0091] b1<b2,h1<h2<h3;

[0092] Score the modeling time to obtain a third score P3. When the modeling time is less than the preset value c1, the third score P3 is the preset value k1. When the modeling time is between the preset values ​​c1 and c2, the third score P3 is the preset value k2. When the modeling time is greater than the preset value c2, the first score P3 is the preset value k3.

[0093] c1<c2,k3<k2<k1;

[0094] Assign a correction value E1 to the first score P1, a correction value E2 to the second score P2, and a correction value E3 to the third score P3;

[0095] E1+E2+E3=1, E3>E2=E1;

[0096] The modeling evaluation information Pp is obtained through the formula P1*E1+P2*E2+P3*E3=Pp;

[0097] The modeling evaluation information of each modeling tool is calculated in sequence, that is, a plurality of modeling evaluation information is obtained.

[0098] The larger the value of the modeling evaluation information Pp is, the better the modeling tool corresponding to the modeling evaluation information is.

[0099] This process comprehensively evaluates the performance of modeling tools by extracting data across multiple dimensions, including processor utilization, memory utilization, storage utilization, and modeling duration. These dimensions collectively reflect the resource utilization and efficiency of the modeling tool during operation, providing a more comprehensive understanding of the tool's performance characteristics.

[0100] By scoring each dimension and calculating a weighted sum, this process provides a comprehensive assessment of the modeling tool's performance. This comprehensive scoring method more accurately reflects the tool's performance in real-world applications, avoiding the one-sidedness of single-dimensional evaluations.

[0101] Adjustable preset values: This process allows for the adjustment of preset values ​​(a1, a2, b1, b2, c1, c2) and the corresponding scores (f1, f2, f3, h1, h2, h3, k1, k2, k3) based on actual circumstances. This means that in practice, the evaluation criteria and weights can be flexibly adjusted based on the specific needs and importance of the modeling tool, resulting in more realistic evaluation results.

[0102] The process provides further flexibility by assigning different correction values ​​(E1, E2, E3) to different scores and allowing the sum of these correction values ​​to be 1. This adjustment can be made according to the emphasis placed on different dimensions in actual applications, making the evaluation results more in line with actual needs.

[0103] This process is based entirely on the modeling tool's objective data (processor utilization, memory utilization, storage device utilization, and modeling time) and pre-set scoring criteria, eliminating the influence of subjective judgment on the evaluation results. This makes the evaluation results more objective and fair.

[0104] Through clear scoring criteria and corresponding preset values, this process ensures the accuracy and consistency of the evaluation process, which helps to maintain the stability and reliability of the evaluation results in practical applications.

[0105] This process can be automated through programming, greatly improving the efficiency of the evaluation. In practical applications, a program can be written to automatically extract relevant data from the modeling tool and calculate the evaluation results based on pre-set scoring criteria and correction values.

[0106] Repeatable Verification: Because the process is data-driven and has clear scoring criteria, its accuracy can be verified by repeatedly collecting data and calculating the evaluation results. This repeatability ensures the stability and reliability of the evaluation results and provides a basis for further optimization and improvement of the modeling tool.

[0107] The optimal model is put into use to monitor clinical data. That is, after the optimal model is put into use, the clinical data collected in real time is imported. When the clinical data collected in real time matches the clinical abnormality status in the optimal model, an alarm message is generated to prompt.

[0108] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "plurality" means at least two, such as two, three, etc., unless otherwise specifically defined.

[0109] In the description of this specification, the reference terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" mean that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0110] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A clinical data monitoring method based on model training, characterized in that: The following steps are involved: Step 1: Data collection, i.e. collecting clinical data, analyzing the data sources during the collection process, and selecting the data sources; Step 2: Perform data preprocessing and obtain preprocessed data; Step 3: Select a modeling method, use the pre-processed data for modeling, collect modeling-related data during the modeling process, and obtain multiple modeling-related data; Step 4: Process multiple modeling-related data to obtain multiple modeling evaluation information, select the best from the multiple modeling evaluation information, and obtain the optimal model; Step 5: Put the optimal model into use and conduct clinical data monitoring.

2. A clinical data monitoring method based on model training according to claim 1, characterized in that: The specific process of step one is as follows: S1: Clarify the data collection objectives, i.e., determine the type of clinical data that needs to be collected; S2: Analyze data sources, evaluate different data sources, and obtain source evaluation information; S3: Analyze the source assessment information and select the final data source for data collection.

3. A clinical data monitoring method based on model training according to claim 2, characterized in that: The process of obtaining the source assessment information: Collect data related to the data source, including information on the number of times the data source is used, data transmission speed, and the number of abnormalities at the data source; The number of times the data source is used is marked as F1, the data transmission speed is marked as F2, and the number of times the data source does not have an abnormality is marked as F3; Assign a correction value M1 to F1, a correction value M2 to F2, and a correction value M3 to F3; M3>M2>M1; Through the formula F1*M1+F2*M2+F3*M3=Ff, the source evaluation information Ff of a single data source is obtained.

4. The clinical data monitoring method based on model training according to claim 3, characterized in that: The specific process of analyzing the source assessment information and selecting the final data source for data collection is as follows: Sort the source evaluation information Ff of a single data source in descending order, then collect the amount of data required for modeling, and select the source corresponding to the source evaluation information Ff as the data source according to the amount of data required for modeling.

5. The clinical data monitoring method based on model training according to claim 1, characterized in that: The specific process of performing data preprocessing and obtaining preprocessed data is as follows: First, perform data cleaning to remove duplicate data, handle missing values ​​and outliers; Use Python’s Pandas library for data cleaning; Then perform data standardization and normalization, standardize the data and scale the data to the required standard range, and then normalize the data and scale the data to a fixed range; After that, data transformation is performed. After necessary transformation of the data, the preprocessed data is obtained.

6. The clinical data monitoring method based on model training according to claim 1, characterized in that: The process of obtaining the modeling-related data is as follows: The preprocessed data is imported into multiple modeling tools for real-time modeling. During the modeling process, the processor usage, memory usage, and modeling time are collected. The processor usage, memory usage, and modeling time constitute the modeling-related data.

7. The clinical data monitoring method based on model training according to claim 6, characterized in that: The process of processing multiple modeling-related data to obtain multiple modeling evaluation information is as follows: Extract the processor usage, memory usage, storage device usage, and modeling time from the modeling-related data of a single modeling tool; Scoring the processor occupancy rate to obtain a first score P1. When the processor occupancy rate is less than a preset value a1, the first score P1 is a preset value f1. When the processor occupancy rate is between a1 and a2, the first score P1 is a preset value f2. When the processor occupancy rate is greater than a2, the first score P1 is a preset value f3. a1<a2,f1<f2<f3; Score the memory usage to obtain a second score P2. When the memory usage is less than the preset value b1, the second score P2 is the preset value h1. When the memory usage is between the preset values ​​b1 and b2, the second score P2 is the preset value h2. When the memory usage is greater than the preset value b2, the second score P2 is the preset value h3. b1<b2,h1<h2<h3; Score the modeling time to obtain a third score P3. When the modeling time is less than the preset value c1, the third score P3 is the preset value k1. When the modeling time is between the preset values ​​c1 and c2, the third score P3 is the preset value k2. When the modeling time is greater than the preset value c2, the first score P3 is the preset value k3. c1<c2,k3<k2<k1; Assign a correction value E1 to the first score P1, a correction value E2 to the second score P2, and a correction value E3 to the third score P3; E1+E2+E3=1, E3>E2=E1; The modeling evaluation information Pp is obtained through the formula P1*E1+P2*E2+P3*E3=Pp; Calculating modeling evaluation information of each modeling tool in sequence, that is, obtaining a plurality of modeling evaluation information; The larger the value of the modeling evaluation information Pp is, the better the modeling tool corresponding to the modeling evaluation information is.

8. The clinical data monitoring method based on model training according to claim 1, characterized in that: The optimal model is put into use to monitor clinical data. That is, after the optimal model is put into use, the clinical data collected in real time is imported. When the clinical data collected in real time matches the clinical abnormality status in the optimal model, an alarm message is generated to prompt.

Images