Information extraction method for clinical test scheme

Abstract

The invention relates to the technical field of clinical test information extraction, and discloses an information extraction method of a clinical test scheme. According to the method, clinical test scheme files, registration information, research papers and related literatures are collected, data are sorted, an electronic database is formed, a text mining technology and a natural language processing tool are used, the collected literatures are analyzed, key information is labeled, and the literatures are analyzed. Calculating a random control test design type proportion and an open label design type proportion, extracting the marked data into a spreadsheet, checking the extracted data, implementing double check, calculating a consistency proportion and an error rate in the checking process, and performing statistical analysis according to the extracted data, so as to obtain a statistical analysis result; calculating a mean value, a standard deviation and a confidence interval, extracting a minimum value and a maximum value of data and a combination effect quantity of a combination end point, arranging analysis results into a report, including detailed description of test design, key results and conclusions, and displaying by using charts and graphs.

Description

Technical Field

[0001] The present invention relates to the technical field of clinical trial information extraction, and in particular to a method for extracting information from a clinical trial protocol. Background Art

[0002] The clinical trial protocol is a systematic description of the entire clinical trial process, which aims to guide researchers on how to conduct a specific clinical trial. It usually includes detailed information such as the background, purpose, design, methods, subject characteristics, sample size, primary and secondary endpoints, and statistical analysis methods of the trial. The clinical trial protocol is an important document to ensure the scientificity, ethics, and repeatability of the trial. The extraction of clinical trial information helps to improve the transparency of the research and ensure that the trial results can be verified and repeated by other researchers, which is crucial for scientific research and even reviewing the safety and effectiveness of new drugs and therapies. Extracting key data from different trials provides the necessary basis for conducting systematic evaluations and meta-analyses, which can integrate evidence from multiple trials and increase the statistical authority of the results. At the same time, medical decision makers, clinicians, and patients can understand the expected effects and risks of new treatment options through the data in the clinical trial protocol to help them make informed decisions.

[0003] Currently, when extracting information from clinical trial protocols, the focus is mainly on a single data source and there is no systematic data collection. As a result, the information data lacks accuracy and reliability when extracting information. Summary of the invention

[0004] 1. Technical issues to be resolved

[0005] In view of the deficiencies of the prior art, the present invention provides a method for extracting information from clinical trial protocols, which has the characteristics of collecting clinical trial protocol documents, registration information, research papers and related literature, organizing data, forming an electronic database, using text mining technology and natural language processing tools, analyzing the collected literature and annotating key information, including the purpose, design, method, subject characteristics, sample size, primary and secondary endpoints and statistical methods of the trial, and calculating the proportion of randomized controlled trial design types and the proportion of open label design types, extracting the annotated data into a spreadsheet, separating the information of the experimental group and the control group, checking the extracted data, implementing double checks, and calculating and checking. The consistency ratio and error rate in the process were calculated, statistical analysis was performed based on the extracted data, sample characteristics were described, mean, standard deviation, confidence interval, minimum and maximum values ​​of the extracted data, and the combined effect size of the combined endpoints were calculated, and the analysis results were compiled into a report, including a detailed description of the experimental design, key results and conclusions, presented using charts and graphs. By comprehensively collecting data from diverse sources such as original literature, registration information, research papers, and forming an electronic database, combined with text mining and natural language processing technology, literature analysis and automatic annotation of key information were performed, information was extracted, and double checks were performed, which improved the accuracy and reliability of information extraction and solved the above-mentioned problems.

[0006] (II) Technical solution

[0007] To achieve the above object, the present invention provides the following technical solution: a method for extracting information of a clinical trial protocol, comprising the following steps:

[0008] S1. Collect clinical trial protocol documents, registration information, research papers and related literature, organize data and form an electronic database;

[0009] S2. Use text mining technology and natural language processing tools to analyze the collected literature and annotate key information, including the purpose, design, methods, subject characteristics, sample size, primary and secondary endpoints, and statistical methods of the trial, and calculate the proportion of randomized controlled trial design types and the proportion of open-label design types;

[0010] S3. Extract the labeled data into a spreadsheet and separate the information of the experimental group and the control group;

[0011] S4. Verify the extracted data, implement double checks, and calculate the consistency ratio and error rate during the verification process;

[0012] S5. Perform statistical analysis based on the extracted data, describe the sample characteristics, and calculate the mean, standard deviation, confidence interval, minimum and maximum values ​​of the extracted data and the combined effect size of the combined endpoints;

[0013] S6. Organize the analysis results into a report that includes a detailed description of the experimental design, key results and conclusions, and uses charts and graphs to present them.

[0014] Preferably, the formula for calculating the proportion of randomized controlled trial design types in S2 is as follows:

[0015]

[0016] In the formula, P rct represents the proportion of randomized controlled trial design types, N Rct represents the number of randomized controlled trials, N total Represents the total number of all trials collected.

[0017] Preferably, the formula for calculating the ratio of open label design types in S2 is as follows:

[0018]

[0019] In the formula, P open represents the proportion of open-label design types, N open represents the number of open-label designs, N total Represents the total number of all trials collected.

[0020] Preferably, the S4 calculates the consistency ratio during the verification process, and the calculation formula is as follows:

[0021]

[0022] In the formula, Ca represents the consistency ratio during the verification process, N yz Indicates the number of consistency checks passed in the double check, N Check Indicates the number of double checks.

[0023] Preferably, the S4 calculates the error rate during the verification process, and the formula is as follows:

[0024]

[0025] In the formula, Et represents the error rate, N er represents the number of errors found in the double check, N Check Indicates the number of double checks.

[0026] Preferably, the S5 calculates the sample feature mean, and the formula is as follows:

[0027]

[0028] In the formula, represents the sample characteristic mean, n represents the sample size, Xt i Represents the i-th sample value, where i represents the counting subscript.

[0029] Preferably, the S5 calculates the sample characteristic standard deviation, and the formula is as follows:

[0030]

[0031] In the formula, St represents the standard deviation of sample characteristics. represents the mean, n represents the sample size, Xt i Represents the i-th sample value, where i represents the counting subscript.

[0032] Preferably, the confidence interval is calculated in S5, and the formula is as follows:

[0033]

[0034] In the formula, CI represents the confidence interval, represents the mean, n represents the sample size, St represents the standard deviation, Represents the critical value of t distribution, and the value range of t is (0-1).

[0035] Preferably, the S5 calculates the minimum and maximum values ​​of the extracted data, and the formula is as follows:

[0036] Zx min =min(Xt1, Xt2, .., Xt m )

[0037] Zx max =max(Xt1, Xt2, .., Xt m )

[0038] In the formula, Zx min With Zx max Indicates the minimum and maximum values ​​of the extracted data, Xt1, Xt2, .., Xt m Represents the m selected sample data.

[0039] Preferably, the S5 calculates the combined effect size of the combined endpoint, and the formula is as follows:

[0040]

[0041] In the formula, Es represents the combined effect size of the combined endpoint. and represent the mean of the experimental group and the control group, St sy With St dz They represent the standard deviation of the experimental group and the control group respectively, and n1 and n2 represent the sample size of the experimental group and the control group respectively.

[0042] Compared with the prior art, the present invention provides a method for extracting information from a clinical trial protocol, which has the following beneficial effects:

[0043] The present invention collects clinical trial protocol documents, registration information, research papers and related documents, organizes data, forms an electronic database, uses text mining technology and natural language processing tools, analyzes the collected documents and annotates key information, including the purpose, design, method, subject characteristics, sample size, primary and secondary endpoints and statistical methods of the trial, and calculates the proportion of randomized controlled trial design types and the proportion of open label design types, extracts the annotated data into a spreadsheet, separates the information of the experimental group and the control group, verifies the extracted data, implements double checks, calculates the consistency ratio and error rate in the verification process, performs statistical analysis based on the extracted data, describes the sample characteristics, calculates the mean, standard deviation, confidence interval, minimum and maximum values ​​of the extracted data and the combined effect size of the combined endpoint, organizes the analysis results into a report, including a detailed description of the trial design, key results and conclusions, and uses charts and graphs for display, comprehensively collects data from diverse sources such as original documents, registration information, research papers, and forms an electronic database, combines text mining and natural language processing technology, performs document analysis and automatic annotation of key information, extracts information, and performs double checks, thereby improving the accuracy and reliability of information extraction. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a schematic diagram of the steps of the method of the present invention. DETAILED DESCRIPTION

[0045] 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.

[0046] In view of the current problem that the information extraction of clinical trial protocols mainly focuses on a single data source and does not systematically collect data, the information data lacks accuracy and reliability during information extraction. Therefore, a method for extracting information from clinical trial protocols is proposed. Figure 1 , the method comprises the following steps:

[0047] S1. Collect clinical trial protocol documents, registration information, research papers and related literature, organize data and form an electronic database;

[0048] In the process of conducting clinical trial research, it is first necessary to comprehensively collect relevant clinical trial protocol documents, registration information, research papers and other relevant literature. This process uses automated data capture tools and web crawler technology to efficiently extract information from major medical databases, clinical trial registration platforms, research journals and professional websites. After obtaining these data, data cleaning and preprocessing techniques are used to remove duplicates and standardize the format to ensure the consistency and accuracy of the information. Next, the database management system is used to store the organized data in a structured form to form an electronic database. This database should have good query performance and scalability to support subsequent data analysis and mining. At the same time, in order to enhance the availability of data, metadata indexes can be established to facilitate rapid retrieval and access to different types of literature information.

[0049] S2. Use text mining technology and natural language processing tools to analyze the collected literature and annotate key information, including the purpose, design, methods, subject characteristics, sample size, primary and secondary endpoints, and statistical methods of the trial, and calculate the proportion of randomized controlled trial design types and the proportion of open-label design types;

[0050] The calculation formula for the proportion of randomized controlled trial design types is as follows:

[0051]

[0052] Randomized controlled trials are one of the standards for evaluating the effectiveness of interventions. By randomly assigning subjects to the experimental group and the control group, selection bias can be effectively reduced, the comparability between groups can be ensured, and the reliability and validity of the research results can be improved. In the formula, P rct represents the proportion of randomized controlled trial design types, N Rct represents the number of randomized controlled trials, N total Represents the total number of all trials collected. Understanding the proportion of randomized controlled trial design types can help researchers assess the prevalence of this design in the extracted literature and thus infer whether the results can be applied to a wider population and different clinical scenarios;

[0053] The calculation formula for the proportion of open-label design types is as follows:

[0054]

[0055] Calculating the proportion of open-label design types can help identify the diversity of such trials in data collection and analysis results, indicating the flexibility and adaptability of clinical research design. In the formula, P open represents the proportion of open-label design types, N open represents the number of open-label designs, N totalRepresenting the total number of all trials collected, identifying the prevalence of open-label designs in specific clinical studies can provide context for subsequent studies, especially in evaluating long-term efficacy or patient experience, where such designs may be more appropriate;

[0056] S3. Extract the labeled data into a spreadsheet and separate the information of the experimental group and the control group;

[0057] After the data is labeled, the labeled data is first systematically extracted into a spreadsheet for subsequent analysis and processing. The data processing library Pandas is used to implement this process to ensure that the data maintains its structural integrity and accuracy when extracted. Next, the information of the experimental group and the control group is separated. With the help of conditional screening and data grouping functions, the relevant characteristics and outcome variables of each group are extracted into different worksheets. This step not only increases the readability of the data, but also simplifies the processing of each group of data in the subsequent analysis process.

[0058] S4. Verify the extracted data, implement double checks, and calculate the consistency ratio and error rate during the verification process;

[0059] The formula for calculating the consistency ratio during the verification process is as follows:

[0060]

[0061] The consistency ratio can be used to measure the consistency between different data sources or annotators. A high consistency ratio indicates that the data is reliable and of high quality, which is crucial for the interpretation of clinical trial results. In the formula, Ca represents the consistency ratio during the verification process, and N yz Indicates the number of consistency checks passed in the double check, N Check Indicates the number of double checks. By calculating the consistency ratio, researchers can obtain a quantitative assessment of the reliability of the data. This transparency can enhance the credibility of the results and help decision makers make more confident judgments on the results of clinical trials.

[0062] The error rate calculation formula is as follows:

[0063]

[0064] Calculating the error rate can provide a quantitative indicator for research to evaluate the types and frequency of errors that occur during data extraction, helping researchers identify and classify the nature of errors. In the formula, Et represents the error rate, N er represents the number of errors found in the double check, N CheckIndicates the number of double checks. By monitoring the error rate, the research team can develop improvement measures to reduce the incidence of errors. This continuous monitoring and improvement mechanism helps to optimize the overall quality of data extraction;

[0065] S5. Perform statistical analysis based on the extracted data, describe the sample characteristics, and calculate the mean, standard deviation, confidence interval, minimum and maximum values ​​of the extracted data and the combined effect size of the combined endpoints;

[0066] The sample feature mean calculation formula is as follows:

[0067]

[0068] The mean is an intuitive expression of the median value in a data set. It can effectively reflect the overall trend of clinical trial results and the average treatment effect of patients, and help researchers quickly understand the distribution of data. In the formula, represents the mean, n represents the sample size, Xt i Represents the i-th sample value, i represents the count subscript, and the calculation of the mean makes the comparison of results between different groups simple and intuitive, especially in randomized controlled trials, which can help evaluate the difference in treatment effects between the experimental group and the control group;

[0069] The formula for calculating the sample characteristic standard deviation is as follows:

[0070]

[0071] The standard deviation is an indicator that measures the distance of the data from the mean. It can evaluate the consistency and variability of the test results. When the standard deviation is small, it means that the data is more consistent. On the contrary, it may indicate that the results vary greatly and the reliability is low. In the formula, St represents the standard deviation. represents the mean, n represents the sample size, Xt i It represents the i-th sample value, i represents the count subscript. By understanding the standard deviation, it can help determine a reasonable sample size during the research design process to enhance the statistical power of the experiment;

[0072] The confidence interval calculation formula is as follows:

[0073]

[0074] The confidence interval can show the uncertainty range of the mean or effect size, quantitatively evaluate the reliability of the results, and enable researchers and clinicians to better understand the range of the results. In the formula, CI represents the confidence interval. represents the mean, n represents the sample size, St represents the standard deviation, It represents the critical value of t distribution. The value range of t is (0-1). The closer to 1, the clearer the sample. The width of the confidence interval can help clinicians more accurately judge the possibility of effect when choosing a treatment plan, which is conducive to clinical decision-making.

[0075] The calculation formulas for the minimum and maximum values ​​of the extracted data are as follows:

[0076] Zx min =min(Xt1, Xt2, .., Xt m )

[0077] Zx max =max(Xt1, Xt2, .., Xt m )

[0078] The minimum and maximum values ​​can intuitively show the range of data variation, help identify extreme values ​​and potential outliers, and provide a basis for data cleaning and analysis. In the formula, Zx min With Zx max Indicates the minimum and maximum values ​​of the extracted data, Xt1, Xt2, .., Xt m Represents the m sample data selected. When analyzing the results, the minimum and maximum values ​​can help researchers evaluate the extreme performance of the samples, so as to determine whether it is necessary to conduct a deeper analysis of the data or adopt different methods to deal with these values;

[0079] The combined effect size of the combined endpoint was calculated as follows:

[0080]

[0081] By calculating the combined effect size of the combined endpoint, researchers can conduct an integrated analysis of multiple important results to obtain a comprehensive evaluation of the treatment effect, which helps to identify the overall benefit of the treatment. In the formula, Es represents the combined effect size of the combined endpoint. and represent the mean of the experimental group and the control group, St sy With St dz Represent the standard deviation of the experimental group and the control group, respectively. n1 and n2 represent the sample size of the experimental group and the control group, respectively. Combining multiple endpoints can improve the statistical power of the study and reduce the possibility of false negative results, especially when the event rate is low. This integration helps to provide stronger evidence support.

[0082] S6. Organize the results of the analysis into a report that includes a detailed description of the experimental design, key results and conclusions, and uses charts and graphs to present them;

[0083] In the analysis report of the clinical trial, a detailed description of the trial design should be fully presented, including key elements such as the study's objectives, inclusion criteria, intervention measures, randomization methods, and blinding design to ensure the reproducibility and scientific nature of the study, and key results should be highlighted. Descriptive statistical analysis and inferential statistics should be used to evaluate the effects of the intervention, and the effect size should be combined to fully present the clinical significance of the results. In order to more intuitively explain these results, charts and graphs, including bar charts, box plots, and survival curves, should be used to clearly display the comparison results and data distribution between different groups, and the corresponding statistical significance levels should be marked. Data visualization tools should be used to transform complex data into easy-to-understand views to enhance the readability and intuitiveness of the report.

[0084] By comprehensively collecting data from diverse sources such as original documents, registration information, research papers, and forming an electronic database, combining text mining and natural language processing technology, conducting literature analysis and automatic annotation of key information, extracting information, and performing double checks, the accuracy and reliability of information extraction are improved.

[0085] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for extracting information from a clinical trial protocol, characterized in that: The following steps are involved: S1. Collect clinical trial protocol documents, registration information, research papers and related literature, organize data and form an electronic database; S2. Use text mining technology and natural language processing tools to analyze the collected literature and annotate key information, including the purpose, design, methods, subject characteristics, sample size, primary and secondary endpoints, and statistical methods of the trial, and calculate the proportion of randomized controlled trial design types and the proportion of open-label design types; S3. Extract the labeled data into a spreadsheet and separate the information of the experimental group and the control group; S4. Verify the extracted data, implement double checks, and calculate the consistency ratio and error rate during the verification process; S5. Perform statistical analysis based on the extracted data, describe the sample characteristics, and calculate the sample characteristic mean, standard deviation, confidence interval, minimum and maximum values ​​of the extracted data, and the combined effect size of the combined endpoints; S6. Organize the analysis results into a report that includes a detailed description of the experimental design, key results and conclusions, and uses charts and graphs to present them.

2. The method for extracting information from a clinical trial protocol according to claim 1, characterized in that: The formula for calculating the proportion of randomized controlled trial design types in S2 is as follows: In the formula, P rct represents the proportion of randomized controlled trial design types, N Rct represents the number of randomized controlled trials, N total Represents the total number of all trials collected.

3. The method for extracting information from a clinical trial protocol according to claim 2, characterized in that: The formula for calculating the proportion of open-label design types in S2 is as follows: In the formula, P open represents the proportion of open-label design types, N open represents the number of open-label designs, N total Represents the total number of all trials collected.

4. The method for extracting information from a clinical trial protocol according to claim 3, characterized in that: The S4 calculates the consistency ratio during the verification process, and the calculation formula is as follows: In the formula, Ca represents the consistency ratio during the verification process, N yz Indicates the number of consistency checks passed in the double check, N Check Indicates the number of double checks.

5. The method for extracting information from a clinical trial protocol according to claim 1, characterized in that: The S4 calculates the error rate during the verification process, and the formula is as follows: In the formula, Et represents the error rate in the verification process, N er represents the number of errors found in the double check, N Check Indicates the number of double checks.

6. The method for extracting information from a clinical trial protocol according to claim 5, characterized in that: The S5 calculates the sample feature mean, and the formula is as follows: In the formula, represents the sample characteristic mean, n represents the sample size, Xt i Represents the i-th sample value, where i represents the counting subscript.

7. The method for extracting information from a clinical trial protocol according to claim 6, characterized in that: The S5 calculates the sample characteristic standard deviation, and the formula is as follows: In the formula, St represents the standard deviation of sample characteristics. represents the mean, n represents the sample size, Xt i Represents the i-th sample value, where i represents the counting subscript.

8. The method for extracting information from a clinical trial protocol according to claim 7, characterized in that: The S5 calculates the confidence interval, and the formula is as follows: In the formula, CI represents the confidence interval, represents the mean, n represents the sample size, St represents the standard deviation, Represents the critical value of t distribution, and the value range of t is (0-1).

9. The method for extracting information from a clinical trial protocol according to claim 8, characterized in that: S5 calculates the minimum and maximum values ​​of the extracted data, and the formula is as follows: Zx min =min(Xt1,Xt2,..,Xt m ) Zx max =max(Xt1,Xt2,..,Xt m ) In the formula, Zx min With Zx max Indicates the minimum and maximum values ​​of the extracted data, Xt1, Xt2, .., Xt m Represents the m selected sample data.

10. The method for extracting information from a clinical trial protocol according to claim 9, characterized in that: The S5 calculates the combined effect size of the combined endpoint, and the formula is as follows: In the formula, Es represents the combined effect size of the combined endpoint. and represent the mean of the experimental group and the control group, St sy With St dz They represent the standard deviation of the experimental group and the control group respectively, and n1 and n2 represent the sample size of the experimental group and the control group respectively.

Images