Clinical test quality control method and system

Abstract

The invention relates to the technical field of data management, in particular to a clinical test quality control method and system. The method comprises the following steps: constructing an initial sample set of a clinical test item based on the number of CRAs in historical clinical test items of the clinical test item and corresponding management data; dividing the initial sample set of the clinical test item into a determined sample and a plurality of to-be-determined sample sets, wherein the determined sample is the same as the clinical test item in category; calculating a calibration index of each to-be-determined sample set; iterating the calibration coefficient and the calibration index of the to-be-determined sample set in a gradient ascending method to obtain a target calibration index of the to-be-determined sample set; calculating a screening index of each to-be-determined sample, wherein the screening index is in positive correlation with the target calibration index; and updating the initial sample set in response to the comparison result of the screening index of each to-be-determined sample and the preset threshold, and obtaining the CRA quantity prediction value, thereby realizing clinical test item quality management, and effectively improving the accuracy of clinical test quality management and control.

Description

Technical Field

[0001] The present invention relates to the technical field of data management, and in particular to a clinical trial quality control method and system. Background Art

[0002] A Clinical Research Associate (CRA) is a professional who is responsible for supervising and managing clinical trials during the pharmaceutical research and development process. In order to ensure the quality of the clinical trial process and the authenticity of clinical trial data, the number of CRA personnel needs to be managed.

[0003] In the prior art, a patent application document with publication number CN114897510A discloses a method and device for human resource adaptation of a clinical trial project. The application obtains the project requirement information of the sponsor of the clinical trial project and the application information of the participants, constructs a first user portrait for the sponsor according to the project requirement information, and constructs a second user portrait for the participant according to the application information; and achieves the adaptation of CRA human resources of the clinical trial project through the matching results of the first user portrait and the second user portrait.

[0004] However, the solutions provided by the above-mentioned prior art do not mention how to accurately obtain the project demand information of the sponsor of the clinical trial project. If the number of CRAs is blindly obtained for matching, there may be an excess or shortage of qualified CRA resources, which will affect the effect of the CRA human resource adaptability of the clinical trial project.

[0005] Based on this, how to accurately obtain and manage the CRA human resource forecast results for clinical trial projects is an urgent problem to be solved by technical personnel in this field. Summary of the invention

[0006] In order to solve the technical problem of how to accurately obtain and manage the CRA human resource forecast results of clinical trial projects, the present invention provides a clinical trial quality control method and system.

[0007] In a first aspect, the present invention provides a clinical trial quality control method, which adopts the following technical solution: A clinical trial quality control method, comprising the steps of: Based on the number of CRAs in the historical clinical trial projects of the clinical trial project and the corresponding management data, the initial sample set of the clinical trial project is constructed; the initial sample set of the clinical trial project is divided into confirmed samples and pending samples, and the pending samples are divided into multiple pending sample sets, where the confirmed samples are of the same category as the clinical trial project; ; is the calibration index of any sample set to be determined, is the size of the pending sample set, To determine the number of samples, is the calibration coefficient of the pending sample set, , Respectively The confirmed samples, the pending sample set The number of CRAs for the pending samples, To manage the number of data types, , Respectively The confirmed samples, the pending sample set The value of the i-th category management data in the pending samples, is an exponential function with base e, is the absolute value symbol; in the gradient ascent method, the calibration coefficient and calibration index of the pending sample set are iterated to obtain the target calibration index of the pending sample set; the screening index of each pending sample is calculated, and the screening index is positively correlated with the target calibration index; in response to the comparison result of the screening index of each pending sample with the preset threshold, the initial sample set is updated to obtain the CRA quantity prediction value, so as to realize the quality management of clinical trial projects.

[0008] The present invention can accurately obtain the predicted value of the number of CRAs required for the current clinical trial project by analyzing the number of CRAs of historical clinical trial projects and the corresponding management data of the current clinical trial project, thereby realizing the quality management of the clinical trial project. In this process, the present invention takes into account that the categories of some historical clinical trial projects are different from the categories of the current clinical trial projects and are inferior samples, but the number of samples with the same category as the current clinical trial project is small; based on this, the present invention can accurately obtain the possibility that the pending samples can be screened as training samples to enrich the sample set by analyzing the proximity between the pending samples with different categories from the current clinical trial project and the determined samples with the same category as the current clinical trial project, so that the sample set of the current clinical trial project can be accurately constructed, and the CRA predicted value of the current clinical trial project can be accurately obtained based on the sample set, effectively improving the accuracy of the quality management of the clinical trial project.

[0009] According to a clinical trial quality control method provided by the present invention, the initial sample set of the clinical trial project is constructed based on the number of CRAs in historical clinical trial projects and corresponding management data of the clinical trial project, including: taking the number of CRAs in each stage of the historical clinical trial project and the corresponding management data as a sample; using the treatment goal of each historical clinical trial project to set a category label for the corresponding sample to obtain the initial sample set.

[0010] The present invention takes into account that the historical clinical trial projects before the current clinical trial project are complex in types and large in number, and therefore sets labels for each type of sample in the initial sample set to facilitate subsequent management and processing of different types of data.

[0011] According to a clinical trial quality control method provided by the present invention, dividing the pending samples into multiple pending sample sets includes: taking pending samples of the same category as one pending sample set.

[0012] According to a clinical trial quality control method provided by the present invention, the calibration coefficient and calibration index of the pending sample set are iterated in the gradient ascent method to obtain the target calibration index of the pending sample set, including: iteratively updating the value of the calibration coefficient of the pending sample set in the gradient ascent method, and obtaining the maximum value of the calibration index in response to a preset iteration termination condition; and using the maximum value of the calibration index as the target calibration index of the pending sample set.

[0013] The present invention takes into account that the calibration coefficient of the pending sample set is used to characterize the degree of similarity between the pending sample set after adjustment and the determined sample. Therefore, the maximum value of the calibration index is obtained in the iterative change of the calibration coefficient, and the degree of similarity between the pending sample set and the determined sample when it is closest to the determined sample after adjustment based on the calibration coefficient is obtained, thereby accurately evaluating the possibility that the pending data set can be used as a training sample.

[0014] According to a clinical trial quality control method provided by the present invention, the calculation of the screening index of each pending sample includes: ; is the screening index for the samples to be determined. is the calibration index of the pending sample set where the pending sample is located, is the size of the pending sample set where the pending sample is located, , are respectively the pending sample and the pending sample set where the pending sample is located. The number of CRAs for the pending samples, To manage the number of data types, , are respectively the pending sample and the pending sample set where the pending sample is located. The i-th category management data value in the pending samples.

[0015] The present invention provides an accurate method for calculating the screening index of pending samples. When the calibration index of the pending sample set in which the pending sample is located is relatively large, the difference between the pending sample and the pending sample set is further verified. The smaller the difference is, the higher the possibility that the pending sample can be used as a training sample is.

[0016] According to a clinical trial quality control method provided by the present invention, the initial sample set is updated in response to the comparison result of the screening index of each pending sample with a preset threshold, including: if the screening index of the pending sample of the clinical trial project is greater than the preset threshold, the pending sample is retained in the initial sample set of the clinical trial project; otherwise, the pending sample is updated and eliminated; and finally the sample set of the clinical trial project is obtained.

[0017] The present invention takes into account that some samples in the initial sample set are low-quality samples, and directly training the prediction model may affect the accuracy of the model's prediction. Therefore, the present invention analyzes the possibility that each pending sample in the initial sample set can be retained in the initial sample set as a training sample, updates the samples of the initial sample set, and finally accurately obtains a high-quality sample set.

[0018] According to a clinical trial quality control method provided by the present invention, the initial sample set is updated in response to the comparison result of the screening index of each pending sample with a preset threshold value to obtain a CRA quantity prediction value to achieve clinical trial project quality management, including: training a random forest model based on the sample set of the clinical trial project to obtain a CRA quantity prediction value; adjusting the CRA quantity of the clinical trial project according to the difference between the CRA quantity of the clinical trial project and the CRA quantity prediction value.

[0019] According to a clinical trial quality control method provided by the present invention, the CRA number of the clinical trial project is adjusted according to the difference between the CRA number of the clinical trial project and the predicted CRA number, including: if the CRA number of the clinical trial project is the same as the predicted CRA number, the CRA number remains unchanged; otherwise, the CRA number of the clinical trial project is adjusted to the predicted CRA number.

[0020] According to a clinical trial quality control method provided by the present invention, the implementation of clinical trial project quality management further includes: generating an analysis report of the clinical trial project based on the CRA quantity and management data corresponding to each stage of the clinical trial project.

[0021] In a second aspect, the present invention provides a clinical trial quality control system, which adopts the following technical solution: A clinical trial quality control system includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned clinical trial quality control method is implemented.

[0022] By adopting the above technical solution, the above-mentioned clinical trial quality control method is generated into a computer program and stored in a memory so as to be loaded and executed by a processor, thereby making a terminal device based on the memory and the processor for easy use.

[0023] The present invention has the following technical effects: Based on the above technical solution, the present invention provides a clinical trial quality control method and system. When managing the CRA human resources of clinical trial projects, by analyzing the number of CRAs in the historical clinical trial projects and the corresponding management data of the current clinical trial projects, the predicted value of the number of CRAs required for the current clinical trial projects can be accurately obtained, thereby realizing the quality management of clinical trial projects. In this process, the present invention takes into account that the categories of some historical clinical trial projects are different from the categories of the current clinical trial projects and are inferior samples, but the number of samples with the same category as the current clinical trial projects is small; based on this, the present invention analyzes the degree of proximity between the pending samples with different categories from the current clinical trial projects and the determined samples with the same category as the current clinical trial projects, and can accurately obtain the possibility that the pending samples can be screened as training samples to enrich the sample set, so that the sample set of the current clinical trial project can be accurately constructed, and the CRA predicted value of the current clinical trial project can be accurately obtained based on the sample set, effectively improving the accuracy of the quality management of clinical trial projects. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A schematic diagram of a process flow in a clinical trial quality control method provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0025] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments.

[0026] In order to accurately obtain and manage the CRA human resources forecast results of clinical trial projects, an embodiment of the present invention discloses a clinical trial quality control method, which obtains the demand forecast value of the number of CRAs by constructing a forecast model, so that CRAs can be reasonably allocated based on the demand forecast value of the number of CRAs.

[0027] For details, please see Figure 1 As shown, Figure 1 A schematic diagram of a process flow in a clinical trial quality control method provided by an embodiment of the present invention, the method specifically comprises the following steps: S1: Construct the initial sample set of the clinical trial project based on the number of CRAs in the historical clinical trial project and the corresponding management data.

[0028] Among them, historical clinical trial projects are projects that have completed complete clinical trials.

[0029] It should be noted that the management data of clinical trial projects may include the weekly enrollment rate of subjects, screening failure rate, subject dropout rate, the number of research centers that CRA needs to manage, the violation rate of research centers, the average daily working hours of CRA, and the intervals between the main stage milestones. The data collection frequency can be set to once a week, that is, the management data of the clinical trial project is counted once a week as a stage.

[0030] The data collection frequency can be set according to actual needs, and the embodiment of the present invention does not impose too many restrictions on this.

[0031] Specifically, the weekly enrollment rate of subjects is the speed at which subjects are screened and successfully enrolled in the research project within this week. If the weekly enrollment rate is lower than the expectation of the clinical trial project, the number of CRAs needs to be increased to promote the recruitment of subjects.

[0032] The screening failure rate is the ratio of the subjects who failed the screening this week to the total number of subjects. Screening failure may be caused by overly strict inclusion criteria or unreasonable screening process. If the screening failure rate is too high, it is necessary to increase CRA investment to optimize the screening process to include more eligible subjects and improve the accuracy and efficiency of screening.

[0033] The subject dropout rate is the ratio of subjects who withdraw this week after joining the clinical trial project at any previous stage. When the subject dropout rate is high, CRA needs to pay close attention to the physical condition and needs of the subjects, so it is necessary to increase CRA follow-up on the subjects to reduce missing data.

[0034] The higher the number of research centers that a CRA needs to manage, the more CRAs are needed to coordinate and manage. The research center violation rate is the total number of times the research center has data missing, plan deviations, etc. this week. The higher the research center violation rate, the more long-term on-site CRA supervision is needed. The average daily working hours of CRAs is one of the indicators that directly reflects the needs of CRAs. If the average daily working hours are too high, it is necessary to adjust the division of labor or increase the number of CRAs in a timely manner.

[0035] Clinical trial projects are usually divided into three main phases, each of which contains multiple phases. The duration of each phase is the same as the acquisition cycle. Milestone nodes are set in each main phase. The interval between the main phase milestone is the number of days between the current acquisition week and the next main phase milestone. The closer to the milestone node, the more CRA short-term intensive monitoring is needed.

[0036] Based on this, the embodiment of the present invention can obtain the number of CRAs and the corresponding management data of the current clinical trial project, build a prediction model by analyzing the number of CRAs and the corresponding management data in the historical clinical trial projects of the current clinical trial project to construct a sample set, input the management data of the current clinical trial project into the prediction model, and obtain the corresponding CRA number prediction value, thereby realizing clinical trial quality management.

[0037] By way of example, in an embodiment of the present invention, an initial sample set of a clinical trial project is constructed based on the number of CRAs in historical clinical trial projects and the corresponding management data of the clinical trial project, including: taking the number of CRAs in each stage of the historical clinical trial projects and the corresponding management data as a sample; using the treatment goals of each historical clinical trial project to set category labels for the corresponding samples to obtain an initial sample set.

[0038] It is understandable that each clinical trial project will set clear treatment goals at the beginning of the trial. Therefore, the embodiment of the present invention marks the clinical trial projects by using the treatment goal categories of the clinical trial projects as labels, so that the clinical trial project samples of various types can be accurately distinguished based on this. For clinical trial projects with multiple treatment goals, at least two labels can be set, which can be set according to actual needs.

[0039] It should be noted that since the current clinical trial project is a research and development project under development, the number of samples with the same type label as the current clinical trial project in the historical clinical trial projects that have completed clinical trials may be small. In order to improve the robustness and accuracy of the prediction model, it is necessary to screen the historical clinical trial project samples in the initial sample set, and remove samples with large differences in the change rules of the current clinical trial project, so as to improve the quality of the sample set and accurately obtain the CRA quantity prediction value, that is, perform the following steps.

[0040] S2: Divide the initial sample set of the clinical trial project into confirmed samples and pending samples, and divide the pending samples into multiple pending sample sets, where the confirmed samples are of the same category as the clinical trial project.

[0041] It should be noted that some of the samples in the initial sample set of the current clinical trial project have the same category labels as the current clinical trial project. Such samples can be directly used as training samples as confirmed samples, but their number is small; although some of the samples have different category labels from the current clinical trial project, if the management data of such samples have similar change patterns to the management data of the confirmed samples, and the difference between the number of CRAs adjusted by the adjustment coefficient and the number of CRAs of the confirmed samples is small, they can also be used as training samples.

[0042] Based on this, the embodiment of the present invention can obtain the calibration coefficient of the pending sample set by analyzing the proximity between the pending sample set of each category and the determined sample, and determine whether the pending sample set can be used as a training sample after correction based on the calibration coefficient.

[0043] It is understandable that the number of pending samples for historical clinical trial projects that have completed clinical trials is relatively large, but the pending samples of the same category have the same treatment goals, so the possibility of using similar management data and CRA numbers in the project implementation process is higher. Therefore, the embodiment of the present invention treats the pending samples of the same category as a pending sample set, and by analyzing the proximity between the pending sample set and the confirmed samples, the data processing volume can be reduced and the data processing efficiency can be improved.

[0044] For example, in the embodiment of the present invention, dividing the pending samples into a plurality of pending sample sets includes: taking the pending samples of the same category as one pending sample set.

[0045] S3: Calculate the calibration index of the pending sample set, iterate the calibration coefficient and calibration index of the pending sample set in the gradient ascent method, and obtain the target calibration index of the pending sample set.

[0046] It should be noted that, among the pending sample sets, the pending sample sets whose management data are closer to the management data of the determined samples are more likely to be used as training samples. If the difference between the numbers of CRAs after adjustment by the calibration coefficient is smaller, the corresponding calibration index will be higher, and the pattern of the number of CRAs in the pending sample set changing with the management data will be closer to the pattern of the number of CRAs in the determined samples changing with the management data.

[0047] Based on this, the embodiment of the present invention can construct a corresponding relationship between the calibration index and the calibration coefficient of the sample set to be determined, and obtain the calibration coefficient corresponding to the maximum calibration index, which is the optimal calibration coefficient. The maximum value of the calibration index is the possibility that the optimal calibration coefficient can be used as a training sample after adjustment.

[0048] For example, in the embodiment of the present invention, the correspondence between the calibration index and the calibration coefficient of the to-be-determined sample set can be specifically referred to the following relationship: ; is the calibration index of any sample set to be determined, is the size of the pending sample set, To determine the number of samples, is the calibration coefficient of the pending sample set, For the The number of CRAs to determine the sample, For the pending sample set The number of CRAs for the pending samples, To manage the number of data types, For the The i-th type of management data value in the determined sample, For the pending sample set The value of the i-th category management data in the pending samples, is an exponential function with base e, is the absolute value symbol.

[0049] Wherein, e is a natural constant. The size of the pending sample set is the number of pending samples contained in the pending sample set.

[0050] In the above formula, the higher the calibration index of the undetermined sample set is, the more similar the change pattern of the undetermined sample set after adjustment based on the calibration coefficient is to the change pattern of the determined sample is, and the higher the possibility that the sample set can be constructed as a training sample is.

[0051] Indicates the first The number of CRAs in the pending samples is the same as that in the The smaller the value is, the smaller the difference in the number of CRAs in the determined samples is. The closer the number of CRAs in the pending samples is to the The number of CRAs to determine the sample, The higher the probability that the pending samples can be used as training samples.

[0052] Indicates the first The management data value of the pending sample is the same as the The smaller the value, the greater the difference between the management data values ​​of the first determined sample. The closer the pending sample is to the The more determined samples there are, the higher the possibility that the undetermined sample can be used as a training sample.

[0053] In order to accurately analyze whether the number of CRA personnel in the pending sample set and the determined sample has the same pattern of change with the management data, it is necessary to make the calibration index as large as possible.

[0054] By way of example, in an embodiment of the present invention, the calibration coefficient and calibration index of the pending sample set are iterated in the gradient ascent method to obtain the target calibration index of the pending sample set, including: iteratively updating the value of the calibration coefficient of the pending sample set in the gradient ascent method, and obtaining the maximum value of the calibration index in response to a preset iteration termination condition; and using the maximum value of the calibration index as the target calibration index of the pending sample set.

[0055] The preset iteration termination condition may be that the number of iterations reaches 100. The preset iteration termination condition and the number of iterations may be set according to actual needs, and the embodiment of the present invention does not impose too many restrictions on this.

[0056] Specifically, in the gradient ascent method, the value of the calibration coefficient of the pending sample set is iteratively updated. In response to the preset iteration termination condition, when the maximum value of the calibration index is obtained, the derivative function of the calibration index of the pending sample set with respect to the calibration coefficient can be used as the gradient function to set the initial value of the calibration coefficient; the initial value of the calibration coefficient is substituted into the gradient function to obtain the gradient value, and a new calibration coefficient is obtained based on the preset learning rate and the gradient value; the new calibration coefficient is used as the initial value to continue to be iteratively updated, and in response to the preset iteration termination condition, the final calibration coefficient is obtained; the calibration index corresponding to the final calibration coefficient is the maximum value of the calibration index.

[0057] Among them, the preset learning rate can be set to 0.01, which can be set specifically according to actual needs; the specific steps of obtaining the maximum value of the calibration index through the gradient ascent method can be implemented by the existing technology, and the embodiments of the present invention are not described in detail here.

[0058] Based on the above steps, the target calibration index of each pending sample set can be obtained. The larger the target calibration index of the pending sample set, the smaller the difference between the pending sample set and the determined sample after adjusting the number of CRAs, the closer the pattern of the change of the number of CRAs with the management data, and the higher the possibility that it can be added to the sample set as a training sample.

[0059] S4: Calculate the screening index of each pending sample, and the screening index is positively correlated with the target calibration index.

[0060] It should be noted that after analyzing the degree of closeness between the pending sample set and the determined sample based on the above steps and obtaining the target calibration index of the pending sample set, the degree of closeness between each pending sample in the pending sample set and the determined sample can be specifically analyzed based on the target calibration index of the pending sample set. The higher the target calibration index of the pending sample set, the higher the credibility of the degree of closeness between each pending sample in the pending sample set and the determined sample.

[0061] For example, in an embodiment of the present invention, the screening index of the pending sample is calculated, and the specific formula may be as follows: ; is the screening index for the samples to be determined. is the calibration index of the pending sample set where the pending sample is located, is the size of the pending sample set where the pending sample is located, is the number of CRAs of the pending sample, is the pending sample set in which the pending sample is located The number of CRAs for the pending samples, To manage the number of data types, The management data value for the i-th category of the pending sample, is the pending sample set in which the pending sample is located The value of the i-th category management data in the pending samples, is an exponential function with base e, is the linear normalization function, is the absolute value symbol.

[0062] Among them, e is a natural constant.

[0063] In the above formula, It indicates the difference between the current pending sample and the pending sample set in which the current pending sample is located. The smaller the value is, the closer the CRA quantity and management data of the current pending sample and the pending sample set in which the current pending sample is located are, and the higher the similarity between the current pending sample and similar pending samples is. Therefore, the possibility that the current pending sample can be used as a training sample is higher, and the corresponding screening index of the current pending sample is higher.

[0064] It indicates the credibility of the possibility that the current pending sample can be used as a training sample. The larger the value, the higher the credibility of the possibility that the current pending sample can be used as a training sample.

[0065] After obtaining the screening index of each pending sample based on the above steps, high-quality training samples can be accurately selected based on the screening index of the pending samples, that is, the following steps are performed.

[0066] S5: In response to the comparison result between the screening index of each pending sample and the preset threshold, the initial sample set is updated to obtain the sample set of the clinical trial project.

[0067] The preset threshold may be set to 0.7; the preset threshold may be set specifically according to actual needs, and the embodiment of the present invention does not impose too many restrictions on this.

[0068] By way of example, in an embodiment of the present invention, the initial sample set is updated in response to the comparison result of the screening index of each pending sample with a preset threshold, including: if the screening index of the pending sample of the clinical trial project is greater than the preset threshold, then the pending sample is retained in the initial sample set of the clinical trial project; otherwise, the pending sample is updated and eliminated; and finally the sample set of the clinical trial project is obtained.

[0069] It can be understood that if the screening index of the pending sample of the current clinical trial project is greater than the preset threshold, it means that the CRA number and the corresponding management data of the pending sample are very close to the CRA number and the corresponding management data of the current clinical trial project. When obtaining the predicted value of the CRA number of the current clinical trial project, the pending sample can be retained as a test sample, and finally a sample set of the current clinical trial project with higher quality is obtained, and a prediction model is constructed based on the sample set, that is, the following steps are executed.

[0070] S6: Implement quality management of clinical trial projects based on the sample sets of clinical trial projects.

[0071] By way of example, in an embodiment of the present invention, the initial sample set is updated in response to the comparison result of the screening index of each pending sample with a preset threshold value to obtain a CRA quantity prediction value to achieve quality management of the clinical trial project, including: training a random forest model based on the sample set of the clinical trial project to obtain a CRA quantity prediction value; and adjusting the CRA quantity of the clinical trial project according to the difference between the CRA quantity of the clinical trial project and the CRA quantity prediction value.

[0072] For example, when training a random forest model based on a sample set of a clinical trial project and obtaining a predicted value for the number of CRAs, the features of the samples can be randomly selected for training, and in response to the preset stop parameters of the random forest, multiple decision trees and corresponding independent prediction results can be obtained, and the parameters can be adjusted to optimize the effect, and finally a prediction model can be obtained; the management data of the clinical trial project can be input into the prediction model to obtain a predicted value for the number of CRAs for the clinical trial project.

[0073] The preset stop parameter of the random forest can be the number of decision trees, the depth of the decision tree, etc., which can be set according to actual needs. The specific steps of training the random forest model through the sample set can be implemented by the existing technology, and the embodiment of the present invention will not be repeated here.

[0074] After obtaining the predicted value of the number of CRAs for the current clinical trial project based on the above steps, the number of CRAs can be adjusted based on the predicted value of the number of CRAs.

[0075] For example, in an embodiment of the present invention, the CRA number of the clinical trial project is adjusted according to the difference between the CRA number of the clinical trial project and the predicted CRA number value, including: if the CRA number of the clinical trial project is the same as the predicted CRA number value, the CRA number remains unchanged; otherwise, the CRA number of the clinical trial project is adjusted to the predicted CRA number value.

[0076] For example, after the number of CRAs for the clinical trial project is adjusted to the predicted value of the number of CRAs, the corresponding number of CRAs can be allocated to the clinical trial project according to the predicted value of the number of CRAs to achieve quality management of the clinical trial project.

[0077] The specific steps of allocating the corresponding number of CRAs to the clinical trial projects can be implemented by the existing technology, and the embodiments of the present invention will not be described in detail here.

[0078] For example, in an embodiment of the present invention, clinical trial project quality management is implemented, and then it also includes: generating an analysis report of the clinical trial project based on the CRA quantity and management data corresponding to each stage of the clinical trial project.

[0079] It can be seen that in the embodiment of the present invention, when implementing clinical trial quality control, the initial sample set of the clinical trial project can be constructed based on the number of CRAs in the historical clinical trial project of the clinical trial project and the corresponding management data; the initial sample set of the clinical trial project is divided into confirmed samples and pending samples, and the pending samples are divided into multiple pending sample sets, where the confirmed samples are of the same category as the clinical trial project; ; is the calibration index of any sample set to be determined, is the size of the pending sample set, To determine the number of samples, is the calibration coefficient of the pending sample set, , Respectively The confirmed samples, the pending sample set The number of CRAs for the pending samples, To manage the number of data types, , Respectively The confirmed samples, the pending sample set The value of the i-th category management data in the pending samples, is an exponential function with base e, is the absolute value symbol; in the gradient ascent method, the calibration coefficient and calibration index of the pending sample set are iterated to obtain the target calibration index of the pending sample set; the screening index of each pending sample is calculated, and the screening index is positively correlated with the target calibration index; in response to the comparison result of the screening index of each pending sample with the preset threshold, the initial sample set is updated to obtain the CRA quantity prediction value, so as to realize the quality management of clinical trial projects and effectively improve the accuracy of clinical trial quality control.

[0080] An embodiment of the present invention further discloses a clinical trial quality control system, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a clinical trial quality control method provided by the present invention is implemented.

[0081] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface, and their configuration and functions are known in the art, so they will not be described in detail here.

[0082] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, apparatus or device.

[0083] The above are all preferred embodiments of the present invention, and are not intended to limit the protection scope of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A clinical trial quality control method, characterized in that: include: Construct an initial sample set for the clinical trial project based on the number of CRAs in the historical clinical trial project and the corresponding management data; Divide the initial sample set of the clinical trial project into confirmed samples and pending samples, and divide the pending samples into multiple pending sample sets, where the confirmed samples are of the same category as the clinical trial project; ； is the calibration index of any sample set to be determined, is the size of the pending sample set, To determine the number of samples, is the calibration coefficient of the pending sample set, , Respectively The confirmed samples, the pending sample set The number of CRAs for the pending samples, To manage the number of data types, , Respectively The confirmed samples, the pending sample set The value of the i-th category management data in the pending samples, is an exponential function with base e, is the absolute value symbol; Iterate the calibration coefficient and calibration index of the undetermined sample set in the gradient ascent method to obtain the target calibration index of the undetermined sample set; The screening index of each pending sample is calculated, and the screening index is positively correlated with the target calibration index; the initial sample set is updated in response to the comparison result of the screening index of each pending sample with the preset threshold value, and the CRA quantity prediction value is obtained to realize the quality management of clinical trial projects.

2. A clinical trial quality control method according to claim 1, characterized in that: The initial sample set of the clinical trial project is constructed based on the number of CRAs in the historical clinical trial project and the corresponding management data, including: The number of CRAs in each stage of historical clinical trial projects and the corresponding management data are taken as a sample; the treatment goals of each historical clinical trial project are used to set category labels for the corresponding samples to obtain the initial sample set.

3. A clinical trial quality control method according to claim 1, characterized in that: The step of dividing the pending samples into a plurality of pending sample sets includes: The pending samples of the same category are regarded as a pending sample set.

4. A clinical trial quality control method according to claim 1, characterized in that: The step of iterating the calibration coefficient and calibration index of the sample set to be determined in the gradient ascent method to obtain the target calibration index of the sample set to be determined includes: The values ​​of the calibration coefficients of the pending sample set are iteratively updated in the gradient ascent method, and a maximum calibration index is obtained in response to a preset iteration termination condition; the maximum calibration index is used as a target calibration index of the pending sample set.

5. A clinical trial quality control method according to claim 1, characterized in that: The step of calculating the screening index of each pending sample includes: ； is the screening index for the samples to be determined. is the calibration index of the pending sample set where the pending sample is located, is the size of the pending sample set where the pending sample is located, , are respectively the pending sample and the pending sample set where the pending sample is located. The number of CRAs for the pending samples, To manage the number of data types, , are respectively the pending sample and the pending sample set where the pending sample is located. The value of the i-th category management data in the pending samples, is a linear normalization function.

6. A clinical trial quality control method according to claim 1, characterized in that: The updating of the initial sample set in response to the comparison result between the screening index of each pending sample and the preset threshold value comprises: If the screening index of the pending sample of the clinical trial project is greater than the preset threshold, the pending sample is retained in the initial sample set of the clinical trial project; otherwise, the pending sample is updated and eliminated; and finally the sample set of the clinical trial project is obtained.

7. A clinical trial quality control method according to claim 6, characterized in that: The initial sample set is updated in response to the comparison result between the screening index of each pending sample and the preset threshold value to obtain the CRA quantity prediction value to achieve the quality management of the clinical trial project, including: The random forest model is trained based on the sample set of the clinical trial project to obtain the predicted value of the number of CRAs; the number of CRAs of the clinical trial project is adjusted according to the difference between the number of CRAs of the clinical trial project and the predicted value of the number of CRAs.

8. A clinical trial quality control method according to claim 7, characterized in that: The adjustment of the number of CRAs for the clinical trial project according to the difference between the number of CRAs for the clinical trial project and the predicted value of the number of CRAs includes: If the number of CRAs for the clinical trial project is the same as the predicted value of CRA number, the number of CRAs remains unchanged; otherwise, the number of CRAs for the clinical trial project is adjusted to the predicted value of CRA number.

9. A clinical trial quality control method according to claim 1, characterized in that: The implementation of clinical trial project quality management also includes: Generate an analysis report on the clinical trial project based on the CRA quantity and management data corresponding to each stage of the clinical trial project.

10. A clinical trial quality control system, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a clinical trial quality control method according to any one of claims 1 to 9 is implemented.

Images