Clinical data mining method, device, electronic device and storage medium
By preset mining rules based on clinical trial stage, recruitment threshold, trial type and end point indicator type, the target clinical trials and data are screened out from a large number of clinical trials, the problem of large amounts and large types is solved, and data mining is improved Efficiency and reliability avoid bias in direction.
Patent Information
- Application Number
- CN202110851809.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-07-27
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2041-07-27
AI Technical Summary
The amount of clinical trial data in the existing technology is huge and the types are large, which leads to a lot of labor and time cost in data research and analysis, and the lack of effective clinical data mining methods, which may lead to biased judgment of mainstream trends.
Provide a clinical data mining method, which can collect clinical trial texts, extract trial data, and screen out target clinical trials and their data based on preset mining rules (including trial stage, recruitment threshold, trial type and end point index type) to avoid Directly use the original data to judge mainstream trends.
Effectively screen out target clinical trials and data for innovative drug clinical development strategies and market competitive landscape analysis, improving data mining efficiency and reliability, avoiding biased direction, and reducing time and labor costs.
Smart Images

Figure CN113674867B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of data mining technology, and in particular to a clinical data mining method, device, electronic equipment and storage medium. Background Art
[0002] The development of innovative drugs requires a large amount of human, material and financial resources to be invested over a long period of time. The clinical development strategy, market competition pattern and project valuation of innovative drugs are of great practical significance. Generally speaking, clinical development strategy, market competition pattern and project valuation require in-depth research on multiple key information, such as epidemiology, clinical guidelines, innovative drug pipelines, clinical trials, clinical data, etc. Among them, clinical trials and clinical data research is a key link that cannot be avoided. At present, clinical trial research can be carried out with the help of official registration platforms such as clinical trials or commercial databases, but these research methods have the following limitations:
[0003] The data recorded on the official registration platform is not comprehensive. For example, there is a lack of information on clinical trial results, and it needs to be obtained from other data sources such as papers, information, etc. The amount of data collected from multiple parties is huge and the data types are diverse. Directly using the collected data for data research and analysis requires a lot of manpower and time costs. In addition, due to the lack of effective mining methods and standards for core clinical data, making judgments on mainstream trends may lead to directional bias. Summary of the invention
[0004] The present invention provides a clinical data mining method, device, electronic device and storage medium, which are used to solve the problem in the prior art that the huge amount of data is not conducive to research and analysis.
[0005] The present invention provides a clinical data mining method, comprising: collecting clinical trial texts; extracting test data of each clinical trial from the clinical trial texts; screening the test data of each clinical trial based on preset mining rules to obtain a target clinical trial and its test data; the preset mining rules are determined based on at least one of the clinical trial stage, the threshold of the number of people recruited for an indication, the clinical trial type, and the endpoint indicator type.
[0006] According to a clinical data mining method provided by the present invention, the preset mining rules include at least one of the following rules: a trial stage mining rule, which is used to mine clinical trials registered and applied for under a preset clinical trial stage; a recruitment number mining rule, which is used to mine clinical trials with a recruitment number greater than a corresponding indication recruitment number threshold; a trial type mining rule, which is used to mine clinical trials under a target trial type; an indicator type mining rule, which is used to mine clinical trials under a target indicator type.
[0007] According to a clinical data mining method provided by the present invention, the test data of each clinical trial is screened based on preset mining rules, including: based on the preset clinical trial stage in the test stage mining rules, and the clinical trial stage and registration declaration status of each clinical trial, the test data of each clinical trial is screened; the clinical trial stage is included in the test data, and the registration declaration status is determined based on the test data or the clinical trial text.
[0008] According to a clinical data mining method provided by the present invention, the registration declaration status is determined based on the relevant announcement text of the corresponding clinical trial, and the relevant announcement text is determined based on the following steps: based on a pre-set association between the drug name and the original research enterprise name, determining the original research enterprise name corresponding to the drug name of any clinical trial; based on the original research enterprise name of any clinical trial and the identification information in the trial data of any clinical trial, extracting the relevant announcement text from the clinical trial text.
[0009] According to a clinical data mining method provided by the present invention, in the recruitment number mining rule, the recruitment number threshold for any indication is determined based on the trial recruitment number of each clinical trial under any indication; the indication and the trial recruitment number are included in the trial data.
[0010] According to a clinical data mining method provided by the present invention, the test data of each clinical trial is screened based on preset mining rules to obtain a target clinical trial and its test data, including: based on the current mining rules, the previous clinical trial set is screened to obtain the current clinical trial set; the current mining rule is determined based on the execution order of the mining rules from the trial stage mining rule, the recruitment number mining rule, the trial type mining rule or the indicator type mining rule, the previous clinical trial set of the first mining rule includes the test data of each clinical trial, and the last clinical trial set includes the target clinical trial and its test data.
[0011] According to a clinical data mining method provided by the present invention, the target clinical trials and their test data are obtained, and then the standardization processing is performed on the test data of each target clinical trial, and the standardization processing includes standardization of at least one of indication information, drug information, sponsor information, test date, test area, number of recruits, test plan, and test status.
[0012] The present invention also provides a clinical data mining device, comprising: a collection unit, an extraction unit and a screening unit. The collection unit is used to collect clinical trial texts; the extraction unit is used to extract the test data of each clinical trial from the clinical trial texts; the screening unit is used to screen the test data of each clinical trial based on preset mining rules to obtain target clinical trials and their test data; the preset mining rules are determined based on at least one of the clinical trial stage, the threshold of the number of people recruited for the indication, the clinical trial type and the endpoint indicator type.
[0013] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of any of the above-mentioned clinical data mining methods are implemented.
[0014] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the clinical data mining methods described above.
[0015] The clinical data mining method, device, electronic device and storage medium provided by the present invention screen out target clinical trials and their test data that are of practical value and can be used for clinical development strategies of innovative drugs and market competition pattern analysis from a large number of clinical trials through preset mining rules determined based on at least one of the clinical trial stage, indication recruitment number threshold, clinical trial type and endpoint indicator type, thereby avoiding the directional bias that may be caused by directly using the original clinical trial data for mainstream trend judgment, and effectively realizing the mining of target clinical data without spending a lot of time and manpower costs, thereby improving data mining efficiency and reliability. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0017] Figure 1 It is a flow chart of the clinical data mining method provided by the present invention;
[0018] Figure 2 is a flowchart of an embodiment of step 130 in the clinical data mining method provided by the present invention;
[0019] Figure 3 is a schematic diagram of the structure of the clinical data mining device provided by the present invention;
[0020] Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION
[0021] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are 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.
[0022] Figure 1 is a flow chart of the clinical data mining method provided by the present invention, such as Figure 1 As shown, the target clinical trial and its test data obtained by the clinical data mining method provided by the embodiment of the present invention can be applied to the formulation of clinical development strategies for innovative drugs, prediction of current and future market competition patterns and project valuation. The method includes the following steps:
[0023] Step 110, collecting clinical trial texts.
[0024] Here, clinical trial text refers to text data containing relevant information of clinical trials. Clinical trial text may specifically include multiple texts, wherein a single text may include text data for information related to a single clinical trial, or may include text data for information related to multiple clinical trials, which is not specifically limited in the embodiments of the present invention. The sources of clinical trial text collection include, but are not limited to: official clinical registration websites of various countries, papers, information, conference abstracts, etc. The collection method may be through crawler collection or search collection, which is not specifically limited in the embodiments of the present invention.
[0025] In implementation, the clinical trial text in step 110 may be collected in real time or periodically according to a preset time interval.
[0026] Step 120, extracting the test data of each clinical trial from the clinical trial text.
[0027] In this step, the test data of each clinical trial is extracted based on the clinical trial. Specifically for drug clinical trials, the test data of each clinical trial includes at least one of the following: clinical stage, clinical registration number, indication information, drug information, inclusion criteria, sponsor information, trial date, trial region, trial plan, and trial status.
[0028] Regarding the extraction of test data, the specific extraction method may be based on keyword matching or rule matching.
[0029] For example, based on keyword matching, "clinical stage" information is extracted from clinical trial texts, from which clinical trials with clinical stages of Phase I, Phase I / II, Phase II, and Phase III are extracted, and the data is cleaned and converted into standard clinical stage data;
[0030] For example, the original data obtained are: Phase 1, Phase 2, and the clinical stages of cleaning standardization correspond to Phase I and Phase II respectively.
[0031] Step 130, based on preset mining rules, the test data of each clinical trial are screened to obtain the target clinical trial and its test data; the preset mining rules are determined based on at least one of the clinical trial stage, the threshold of the number of people recruited for the indication, the clinical trial type and the endpoint indicator type.
[0032] Here, the target clinical trial is the clinical trial that needs to be mined and screened, and the preset mining rules of the target clinical trial can be specified by the user. Considering that core clinical trials are of guiding significance for the clinical development strategy of innovative drugs, the prediction of current and future market competition patterns and project valuation, core clinical trials can be used as target clinical trials for data mining. Here, the target clinical trial must meet at least one of the following two criteria:
[0033] (1) Registration clinical trials initiated by developers, which can support regulatory approval;
[0034] (2) High-quality, investigator-initiated randomized controlled clinical trials that can support changes in clinical guidelines.
[0035] In order to obtain target clinical trials that meet the above two criteria, standardized preset mining rules need to be established. The setting of the preset mining rules here takes into account at least one of the four dimensions including clinical trial stage, indication recruitment threshold, clinical trial type and endpoint indicator type.
[0036] Among them, based on the preset mining rules determined in the clinical trial stage, starting from the differences in the test objectives of different clinical stages, different mining rules are set for different clinical trial stages. For example, the clinical stage can be Phase I, Phase I / II, Phase II or Phase III. Phase I clinical trials include preliminary clinical pharmacology, human safety evaluation tests and pharmacokinetic tests, which provide a basis for formulating dosing regimens. The purpose of Phase II clinical trials is to preliminarily evaluate the therapeutic effect and safety of drugs on patients with target indications, and also to provide a basis for the design of Phase III clinical trials and the determination of dosing regimens. The purpose of Phase III clinical trials is to further verify the therapeutic effect and safety of drugs on patients with target indications, evaluate the relationship between benefits and risks, and ultimately provide sufficient basis for the review of drug registration applications. Based on the test objectives of the above clinical trial stages, Phase III clinical trials can be set as clinical trials that need to be mined, and Phase I, Phase I / II or Phase II clinical trials can also be set to combine other information to determine whether they are clinical trials that need to be mined.
[0037] Based on the preset mining rules determined by the threshold of the number of people recruited for the indication, we can start from the epidemiological status of different indications, set different ranges of the number of people recruited for different indications, and combine the actual number of people recruited in the clinical trial to determine whether the clinical trial is a clinical trial that needs to be mined. Here, the threshold of the number of people recruited for the indication can be obtained through data analysis or pre-set by the staff. For the setting of the number of people recruited for different indications, the epidemiological status of the indication can be specifically considered. The larger the epidemiological population, the more people will be recruited in the clinical trial. The higher the credibility of the clinical trial that meets the recruitment number, the more it can support regulatory approval and support changes in clinical guidelines.
[0038] Based on the preset mining rules determined by the clinical trial type, the clinical trial types include exploratory, prospective, retrospective, registered clinical, post-marketing research or real-world research, etc., the mining rules can be set according to the type of target clinical trial that needs to be mined. For example, the target clinical trial can be set as a registered clinical type, and clinical trials belonging to this type can be mined; or the type excluded from the target can be set to filter out clinical trials belonging to this type, such as excluding open-label extended clinical trials, retrospective studies, and exploratory trials of dosage and medication sequence from the target clinical trial.
[0039] Based on the preset mining rules determined by the endpoint indicator type, the endpoint indicator type includes the primary endpoint indicator or the secondary endpoint indicator. The mining rules can be set according to the endpoint indicator type of the target clinical trial that needs to be mined. For example, the target clinical trial endpoint indicator type can be set as the primary endpoint indicator as an indicator related to effectiveness, and the clinical trials belonging to this type can be mined; or the type excluded from the target can be set to filter out the clinical trials belonging to this type, such as filtering out clinical trials with safety-related indicators.
[0040] By applying preset mining rules and conducting target clinical trial mining based on clinical trials, we can obtain target clinical trials that can meet actual mining needs, as well as the experimental data of each target clinical trial.
[0041] The method provided in the embodiment of the present invention screens out target clinical trials and their test data that are of practical value for clinical development strategies of innovative drugs and market competition pattern analysis from a large number of clinical trials by using preset mining rules determined based on at least one of the clinical trial stage, the threshold of the number of people recruited for the indication, the clinical trial type, and the endpoint indicator type, thereby avoiding the directional bias that may be caused by directly using the original clinical trial data for mainstream trend judgment, and effectively realizing the mining of target clinical data without spending a lot of time and manpower costs, thereby improving data mining efficiency and reliability.
[0042] Based on the above embodiment, step 110 includes:
[0043] Clinical trial texts are collected based on official clinical registration websites, papers, information, and conference abstracts of various countries.
[0044] Here, by collecting clinical trial texts from various data sources, the sources of clinical trial data are richer than relying solely on official registration platforms or commercial databases, ensuring that the data is comprehensive and without omissions. At the same time, the clinical trial data from various data sources are integrated into one platform to facilitate subsequent data mining and analysis research.
[0045] Based on the above embodiment, the preset mining rules include at least one of the following rules:
[0046] Trial stage mining rules are used to mine clinical trials registered and submitted under the preset clinical trial stage;
[0047] The recruitment number mining rule is used to mine clinical trials with recruitment numbers greater than the corresponding indication recruitment number threshold;
[0048] Trial type mining rules, used to mine clinical trials under the target trial type;
[0049] Indicator type mining rules are used to mine clinical trials under the target indicator type.
[0050] Specifically, the trial phase mining rule is used to screen out target clinical trials that meet the trial phase requirements for registration and application. In the trial phase mining rule, the preset clinical trial phase is one or more pre-set clinical trial phases. For example, common clinical trial phases include Phase I, Phase I / II, Phase II, and Phase III, and one or more of them can be used as preset clinical trial phases.
[0051] For example, when the preset clinical trial stage is Phase III, considering that Phase III originally provides the basis for the review of drug registration applications, it can be assumed that Phase III is eligible for registration declaration. If the stage of any clinical trial is Phase III, the clinical trial can be directly identified as a clinical trial that needs to be explored.
[0052] For another example, when the preset stage is other Phase I, Phase I / II or Phase II, it is possible to determine whether the clinical trial is used for registration declaration based on the information contained in the test data, or based on the relevant announcement text of the corresponding clinical trial and the content related to registration declaration in the announcement text to determine whether it is used for registration declaration. If both conditions are met, it is determined that the clinical trial is a clinical trial that needs to be mined.
[0053] The recruitment number mining rule is used to mine clinical trials whose recruitment number is greater than the recruitment number threshold for the corresponding indication.
[0054] Here, the indications and recruitment thresholds of the clinical trial need to be determined through test data. In order to ensure the effectiveness of the clinical trial, it is necessary to determine that the number of people recruited for the corresponding indication of the clinical trial is greater than the corresponding threshold of the number of people recruited for the indication. The corresponding threshold of the number of people recruited for the indication can be obtained through data analysis or preset. If the number of people recruited for the clinical trial is greater than the corresponding threshold of the number of people recruited for the indication, it is confirmed that the clinical trial belongs to the target clinical trial.
[0055] The trial type mining rule is used to mine clinical trials under the target trial type. In one embodiment, the specific mining process is as follows:
[0056] (1) Label clinical trials by trial type
[0057] Obtain the clinical trial protocol corresponding to each clinical trial, and bind the clinical trial protocol to the trial type label through entity recognition or rule-based matching; here, the trial type label can be "prospective", "non-blind (open)", "double-blind", "multi-center", "single-arm", "placebo-controlled", etc.;
[0058] (2) Screening clinical trials based on trial type labels
[0059] For example, the trial type labels "double-blind", "placebo-controlled" and "single-arm" can be pre-set as target trial types, and clinical trials with trial types of "double-blind", "placebo-controlled" or "single-arm" can be screened from various clinical trials as target clinical trials.
[0060] Indicator type mining rules are used to mine clinical trials under the target indicator type. The target indicator type here can be set to the primary endpoint indicator related to effectiveness. Obtain the clinical trial plan corresponding to each clinical trial, and obtain the clinical primary endpoint indicator through entity recognition or rule-based matching. If the primary endpoint indicator is an indicator related to effectiveness (not safety-related indicator), then include the clinical trial in the target clinical trial.
[0061] The effectiveness-related indicators may include: OS (Overall Survival), PFS (progression-free survival), DFS (Disease-free survival), ORR (Objective Response Rate), DCR (Disease control rate); for example, the clinical trial protocol contained in the test data of any clinical trial can be input into the entity recognition model in the form of text to obtain the indicator type output by the entity recognition model as an entity, as shown in the following table, where the text is the clinical trial protocol, and the primary endpoint is the indicator type obtained by entity recognition. If the primary endpoint PFS is an effectiveness endpoint indicator, then the clinical trial belongs to the target clinical trial.
[0062]
[0063] Based on the above embodiment, step 130 includes:
[0064] Based on the preset clinical trial stages in the trial stage mining rules, as well as the clinical trial stages and registration status of each clinical trial, the trial data of each clinical trial are screened; the clinical trial stage is included in the trial data, and the registration status is determined based on the trial data or clinical trial text.
[0065] The registration application status is determined based on the test data or clinical trial text, and can be obtained in one or more of the following ways:
[0066] Obtained from background information of clinical results related materials such as papers and conference materials: Background information of clinical results related materials such as papers and conference materials usually indicates whether the clinical trial is used for registration application. Clinical results related materials are included in the clinical trial text;
[0067] Obtain from drug instructions: In some clinical research information, the drug instructions are associated with the clinical registration number. The drug instructions can be obtained through the clinical registration number contained in the test data, and then the registration application status can be obtained;
[0068] Obtained from review materials: When applying for marketing, the drug will submit review materials to the U.S. Food and Drug Administration (FDA). Usually, the review materials are associated with the clinical registration number. The review materials can be obtained through the clinical registration number contained in the test data, and then the registration application status can be obtained;
[0069] Obtained from company announcements: If the clinical trial stage of any clinical trial meets the preset clinical trial stage set in the trial stage mining rules, the clinical trial can be traced through the drug name contained in the trial data of the clinical trial, thereby obtaining the relevant announcement text of the clinical trial, and determining the registration status of the clinical trial accordingly, so as to combine the clinical trial stage and registration status of the clinical trial to determine whether the clinical trial may be a target clinical trial.
[0070] Furthermore, the registration application status is determined based on the relevant announcement text of the corresponding clinical trial, and the relevant announcement text is determined based on the following steps:
[0071] Based on the pre-set association between the drug name and the original research company name, determine the original research company name corresponding to the drug name of any clinical trial; based on the original research company name of the clinical trial and the identification information in the trial data of the clinical trial, extract the relevant announcement text from the clinical trial text.
[0072] Here, the relevant announcement text may be an announcement related to clinical trials issued by the original drug research enterprise, or it may be a drug clinical trial approval notice issued by the State Drug Administration or other announcements related to clinical trials. The embodiment of the present invention does not specifically limit this.
[0073] For example, under the premise that the clinical trial stage of any clinical trial is the preset clinical trial stage, the relevant announcement text can be extracted from the clinical trial text based on the drug name of the corresponding clinical trial using the keyword matching method, and combined with the content related to registration application in the relevant announcement text, it is determined whether the clinical trial is used for registration application. If it is determined to be used for registration application, then the clinical trial is a clinical trial that needs to be mined.
[0074] Specifically, the association between the drug name and the original research company name is pre-set, and the original research company name corresponding to the drug name can be obtained by querying the drug name of any clinical trial.
[0075] On this basis, the relevant announcement text can be extracted in combination with the name of the original research enterprise of the clinical trial and the identification information in the test data of the clinical trial. Here, the identification information is the identifier of the clinical trial, which can be a clinical registration number or a clinical stage + indication + drug name. For example, you can first extract the announcement issued by the original research enterprise from the clinical trial text, and then filter out the target announcement related to the clinical trial through the clinical registration number, or you can directly filter out the relevant announcement text from the clinical trial text in the form of clinical stage + drug name + indication. The embodiment of the present invention does not specifically limit this.
[0076] In one embodiment, the screening process based on the mining rules in the trial phase is as follows:
[0077] (1) Obtain “clinical stage” information from clinical trial texts and extract clinical trials with phase I, phase I / II, phase II, and phase III clinical trials;
[0078] (2) Extract clinical registration numbers, drug information, and indication information from the clinical trials obtained above to obtain standardized drug names;
[0079] (3) Obtain the name of the original research company from the established drug-original research company relationship and obtain the original research company announcement from the clinical trial text.
[0080] (4) Screen out announcements related to clinical trials from original research announcements and determine whether the announcement confirms that the clinical trial is for registration application;
[0081] For example, the target announcements can be screened out by drug name + indication + clinical stage, or by clinical registration number;
[0082] For example, in the announcement issued by the target original research enterprise, the target announcement information is screened out through "Cadonilimab" + "metastatic cervical cancer" + "Phase II" as "In January 2021, Cadonilimab completed the enrollment of patients in the registration Phase II clinical trial for the treatment of recurrent or metastatic cervical cancer in China."
[0083] Based on the above information, it is determined that the Phase II clinical trial of the drug under this indication is used for registration application. Therefore, this clinical trial is included in the target clinical trial.
[0084] Based on the above embodiments, the recruitment threshold for any indication is determined based on the number of trial recruits for each clinical trial under the indication.
[0085] Specifically, the number of trial recruits for the indication can be obtained through the trial data of any clinical trial, and the number of trial recruits for different clinical trials with the same indication can be aggregated according to the same indication to obtain the range of the number of trial recruits for the clinical trials under each indication, and thereby determine the threshold number of recruits for each indication.
[0086] It should be noted that the indications referred to here may be sub-indications.
[0087] In one embodiment, the specific mining process is as follows:
[0088] (1) Pre-establish a dictionary of segmented indications
[0089] Obtain the indication information and inclusion criteria corresponding to each clinical registration number from the clinical trial text, and identify in real time from the above two information, including but not limited to the following dimensions: ① disease name, ② disease stage; ③ biomarkers; ④ therapy positioning; ⑤ therapy type; ⑥ previous treatment plan; ⑦ patient baseline (such as patient age, gender, etc.). Based on the above dimensions, comprehensively construct a set of segmented indication dictionaries that meet market valuation requirements;
[0090] For example, the subdivided indications of non-small cell lung cancer are as follows:
[0091] Non-small cell lung cancer / advanced stage / EGFR mutation / first-line
[0092] Non-small cell lung cancer / advanced stage / EGFR mutation / second-line
[0093] Non-small cell lung cancer / stage II-IIIB / EGFR mutation / resectable / neoadjuvant
[0094] (2) Obtaining the recruitment threshold corresponding to the segmented indications
[0095] Based on the above steps, obtain the subdivided indications and the number of recruits for each clinical trial registration number from the clinical trial text, aggregate the data according to the same subdivided indications, obtain the range of the number of recruits for each indication through data analysis, and thereby determine the threshold number of recruits for each indication;
[0096] (3) Determine whether the number of clinical trial recruits meets the requirements
[0097] Obtain the subdivided indications and the number of participants recruited for each clinical trial. If the number of participants meets the requirements, the clinical trial will be included.
[0098] For example, through big data analysis, we know that the number of recruits for non-small cell lung cancer / advanced stage / EGFR mutation / first-line is ≥150, and 150 can be used as the threshold for recruiting, that is, the threshold for judging whether the number of recruits for the trial is reasonable. If the number of recruits for the above clinical trial with the indication of non-small cell lung cancer / advanced stage / EGFR mutation / first-line is significantly less than 150 (such as only 50), the clinical trial will be excluded.
[0099] Based on any of the above embodiments, step 130 includes:
[0100] Based on the current mining rules, the previous clinical trial set is screened to obtain the current clinical trial set; the current mining rules are determined based on the execution order of the mining rules, from the trial stage mining rules, the recruitment number mining rules, the trial type mining rules or the indicator type mining rules. The previous clinical trial set of the first mining rule includes the trial data of each clinical trial, and the last clinical trial set includes the target clinical trial and its trial data.
[0101] Specifically, considering the huge amount of clinical trial data, in order to further improve the efficiency of data screening, the above four preset mining rules can be used in sequence to perform progressive stratified screening on the test data of each clinical trial.
[0102] Here, the execution order of the mining rules can be determined according to the needs of the user, and the embodiment of the present invention does not make any specific limitation on this.
[0103] In a specific example, Figure 2 As shown, the mining process is as follows:
[0104] Step 210, based on the trial phase mining rule, the trial data of each clinical trial is screened to obtain a first clinical trial set;
[0105] Step 220, based on the recruitment number mining rule, the first clinical trial set is screened to obtain a second clinical trial set;
[0106] Step 230, screening the second clinical trial set based on the trial type mining rule to obtain a third clinical trial set;
[0107] Step 240 , based on the indicator type mining rules, the third clinical trial set is screened to obtain the target clinical trial and its trial data.
[0108] After each screening of the preset mining rules, a candidate set is obtained, and the target clinical trial search space is reduced. After four mining screenings, the target clinical trial and its test data are finally obtained, thereby maximizing the amount of data and improving the execution speed and efficiency of data mining.
[0109] It should be noted that the first clinical trial set obtained in step 210 is a set of clinical trials and their test data that meet the preset trial stage mining rules; on this basis, based on the recruitment number mining rules, the second clinical trial set mined is a set of clinical trials and their test data that meet the preset trial stage mining rules and the recruitment number mining rules; on this basis, based on the trial type mining rules, the third clinical trial set mined is a set of clinical trials and their test data that meet the preset trial stage mining rules, the recruitment number mining rules and the experiment type mining rules; on this basis, based on the indicator type mining rules, the target clinical trials and their test data mined are clinical trials and their test data that meet the preset trial stage mining rules, the recruitment number mining rules, the experiment type mining rules and the indicator type mining rules.
[0110] Based on any of the above embodiments, the target clinical trial and its trial data are obtained, and then the method also includes: standardizing the trial data of each target clinical trial, and the standardization includes standardization of at least one of the indication information, drug information, sponsor information, trial date, trial area, number of recruits, trial plan, and trial status.
[0111] Specifically, standardized clinical trial related data include but are not limited to the following:
[0112] (1) Standardized indication information
[0113] The disease information corresponding to each clinical registration number is obtained from the trial data of the target clinical trial, and matched with the established indication dictionary to obtain the standard indication name.
[0114] like,
[0115] Raw data Standardizing Data Non Small Cell Lung Cancer Non-small cell lung cancer
[0116] (2) Standardize drug information and obtain target information from it
[0117] Obtain the treatment plan information corresponding to each clinical registration number from the test data of the target clinical trial, extract the clinical trial drug name through entity recognition, match it with the established drug dictionary in the database, and obtain the standard drug name;
[0118] Obtain the target name corresponding to the drug name based on the established association between the drug and the target;
[0119] (3) Information of the standardization sponsor
[0120] Obtain the sponsor information corresponding to each clinical registration number from the trial data of the target clinical trial, match it with the enterprise dictionary established in the database, and obtain the standard sponsor information;
[0121] (4) Standardized clinical trial start date, completion date, and clinical trial days
[0122] Obtain the clinical trial start date and end date information corresponding to each clinical registration number from the test data of the target clinical trial, and convert the data into a standard date format after cleaning; for example,
[0123] Raw data:
[0124] Actual Study Start Date:September 16,2015
[0125] Actual Primary Completion Date:September 15,2017
[0126] Clean the transformed data:
[0127] Start date: 2015-09-16
[0128] End Date: 2017-09-15
[0129] Clinical trial days (days) = end date - start date;
[0130] (5) Standardized test area
[0131] Obtain the experimental location information corresponding to each clinical registration number from the experimental data of the target clinical trial, extract the country information from the data, and convert the country name into the country abbreviation; for example,
[0132] The original data is "China, Beijing", and the cleaned data is "CN";
[0133] The original data is "New York, New York, United States", and the cleaned data is "US";
[0134] (6) Number of clinical recruits
[0135] Obtain the number of recruits corresponding to each clinical registration number from the trial data of the target clinical trial, and extract digital information from it; for example,
[0136] The original data is: 450 participants, so the number of clinical recruitment is: 450
[0137] (7) Standardized test protocol
[0138] Obtain the treatment plan information and control plan information corresponding to each clinical registration number from the test data of the target clinical trial, identify the drug name from the above information, match it in the established drug dictionary in the database, and obtain the standard test group drug name and control group drug name; for example,
[0139] The treatment regimen obtained from the clinical registration number NCT03052608 was standardized and the results were as follows:
[0140] disease Experimental Group Control group Non-small cell lung cancer / advanced stage / ALK positive / first-line Lorlatinib Crizotinib
[0141] (8) Formatting clinical trial results information
[0142] Obtain the experimental methods and experimental results corresponding to each clinical registration number from the experimental data of the target clinical trial,
[0143] ① Obtain the preset clinical primary / secondary endpoint indicators from the trial protocol through entity recognition or rule-based matching, such as:
[0144]
[0145]
[0146] ②According to the above primary / secondary endpoint information, obtain the "primary endpoint data" or "secondary endpoint data" or "adverse reaction data" corresponding to the target clinical trial from the corresponding test results, format them, and extract the endpoint indicators and corresponding data. For example,
[0147]
[0148] (9) Standardized test status
[0149] Pre-defined test status is standardized into 3 states: in progress, completed, and terminated
[0150] Extract the test status information from the test data of the target clinical trial, and based on the extracted test status information, convert the test status into three standard simple states; such as,
[0151] ① If the information obtained is "Completed", mark it as "Completed";
[0152] ② If the information obtained is "Suspended" or "Terminated" or "Withdrawn", it will be marked as "Terminated";
[0153] ③ If the information obtained is "Not yet recruiting" or "Recruiting" or "Enrolling by invitation" or "Active, not recruiting", then determine whether the above clinical result information exists. If so, mark it as "completed"; if not, mark it as "in progress."
[0154] It should be noted that the above standardization scheme can also be applied to the cleaning and standardization processing of the test data of each clinical trial obtained in step 120, thereby obtaining standardized test data of each clinical trial to provide support and convenience for subsequent data mining.
[0155] By adopting the method of the embodiment of the present invention, drugs, indications, clinical data and clinical trial registration information are highly standardized, and ultimately an association relationship between indications, drugs, clinical results and clinical trial registration information is established, which can quickly perform field trend analysis for specific targets or diseases.
[0156] In addition, after completing the standardized processing of the test data of each target clinical trial, an efficient core clinical data analysis and mining platform can be formed.
[0157] The clinical data mining device provided by the present invention is described below. The clinical data mining device described below and the clinical data mining method described above can be referenced to each other.
[0158] Figure 3 is a schematic diagram of the structure of the clinical data mining device provided by the present invention, such as Figure 3 As shown, the device comprises:
[0159] A collection unit 310, used for collecting clinical trial texts;
[0160] An extraction unit 320, configured to extract the test data of each clinical trial from the clinical trial text;
[0161] The screening unit 330 is used to screen the test data of each clinical trial based on preset mining rules to obtain the target clinical trial and its test data; the preset mining rules are determined based on at least one of the clinical trial stage, the threshold of the number of people recruited for the indication, the clinical trial type and the endpoint indicator type.
[0162] The device provided by the embodiment of the present invention collects clinical trial texts from various data sources and extracts the test data of each clinical trial from the clinical trial text. Compared with relying solely on official registration platforms or commercial databases, the sources of test data for clinical trials are richer, ensuring that the data are comprehensive and not missed. At the same time, the clinical trial data from various data sources are centrally integrated to facilitate later data mining and analysis research; through preset mining rules determined based on at least one of the clinical trial stage, the threshold of the number of people recruited for the indication, the clinical trial type and the endpoint indicator type, target clinical trials and their test data that are of practical value and can be used for the clinical development strategy of innovative drugs and the analysis of the market competition landscape are screened out from a large number of clinical trials registered globally, avoiding the directional bias that may be caused by directly using the original clinical trial data for mainstream trend judgment.
[0163] Based on the above embodiment, the screening unit 330 is used for:
[0164] Based on the preset clinical trial stage in the trial stage mining rule, as well as the clinical trial stage and registration declaration status of each clinical trial, the trial data of each clinical trial is screened; the clinical trial stage is included in the trial data, and the registration declaration status is determined based on the trial data or the clinical trial text.
[0165] Based on any of the above embodiments, the screening unit 330 is used to:
[0166] Based on the current mining rules, the previous clinical trial set is screened to obtain the current clinical trial set;
[0167] The current mining rule is determined based on the execution order of the mining rules, from the trial stage mining rule, the recruitment number mining rule, the trial type mining rule or the indicator type mining rule. The previous clinical trial set of the first mining rule includes the trial data of each clinical trial, and the last clinical trial set includes the target clinical trial and its trial data.
[0168] Figure 4 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 4As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430 and a communication bus 440, wherein the processor 410, the communication interface 420 and the memory 430 communicate with each other through the communication bus 440. The processor 410 may call the logic instructions in the memory 430 to execute the clinical data mining method, which includes: collecting clinical trial texts; extracting the test data of each clinical trial from the clinical trial texts; screening the test data of each clinical trial based on preset mining rules to obtain the target clinical trial and its test data; the preset mining rules are determined based on at least one of the clinical trial stage, the threshold of the number of people recruited for the indication, the clinical trial type and the endpoint indicator type.
[0169] In addition, the logic instructions in the above-mentioned memory 430 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.
[0170] On the other hand, the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the clinical data mining method provided by the above-mentioned methods, the method including: collecting clinical trial texts; extracting trial data of each clinical trial from the clinical trial texts; based on preset mining rules, screening the trial data of each clinical trial to obtain a target clinical trial and its trial data; the preset mining rules are determined based on at least one of the clinical trial stage, the indication recruitment number threshold, the clinical trial type and the endpoint indicator type.
[0171] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the above-mentioned clinical data mining methods, the methods comprising: collecting clinical trial texts; extracting trial data of each clinical trial from the clinical trial texts; based on preset mining rules, screening the trial data of each clinical trial to obtain a target clinical trial and its trial data; the preset mining rules are determined based on at least one of the clinical trial stage, the threshold number of people recruited for the indication, the clinical trial type, and the endpoint indicator type.
[0172] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.
[0173] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.
[0174] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A clinical data mining method, It is characterized in that include: Collect clinical trial texts; Extracting the trial data of each clinical trial from the clinical trial text; Based on preset mining rules, the test data of each clinical trial are screened to obtain a target clinical trial and its test data; the preset mining rules are determined based on at least one of the clinical trial stage, the threshold of the number of people recruited for the indication, the clinical trial type and the endpoint indicator type; the clinical trial type and the endpoint indicator type are obtained by entity recognition or rule-based matching for the test scheme in the test data; The target clinical trial includes a core clinical trial, which includes a registration clinical trial initiated by a developer and / or a high-quality randomized controlled clinical trial initiated by a researcher; the preset mining rules include at least one of the following rules: Trial stage mining rules are used to mine clinical trials registered and submitted under the preset clinical trial stage; The recruitment number mining rule is used to mine clinical trials with recruitment numbers greater than the corresponding indication recruitment number threshold; Trial type mining rules, used to mine clinical trials under the target trial type; The indicator type mining rule is used to mine clinical trials under the target indicator type, and the target indicator type includes the main endpoint indicator related to effectiveness.
2. The clinical data mining method according to claim 1, It is characterized in that The screening of the test data of each clinical trial based on the preset mining rules includes: Based on the preset clinical trial stage in the trial stage mining rule, and the clinical trial stage and registration declaration status of each clinical trial, the trial data of each clinical trial are screened; The clinical trial stage is included in the trial data, and the registration declaration status is determined based on the trial data or the clinical trial text.
3. The clinical data mining method according to claim 2, It is characterized in that The registration application status is determined based on the relevant announcement text of the corresponding clinical trial, and the relevant announcement text is determined based on the following steps: Based on the pre-set association between drug names and original research company names, determine the original research company name corresponding to the drug name of any clinical trial; Based on the name of the original research enterprise of any clinical trial and the identification information in the trial data of any clinical trial, the relevant announcement text is extracted from the clinical trial text.
4. The clinical data mining method according to claim 1, It is characterized in that In the recruitment number mining rule, the recruitment number threshold for any indication is determined based on the number of trial recruits in each clinical trial under the any indication; the indication and the number of trial recruits are included in the trial data.
5. The clinical data mining method according to any one of claims 1 to 4, It is characterized in that The method of screening the test data of each clinical trial based on the preset mining rules to obtain the target clinical trial and its test data includes: Based on the current mining rules, the previous clinical trial set is screened to obtain the current clinical trial set; The current mining rule is determined based on the execution order of the mining rules, from the trial stage mining rule, the recruitment number mining rule, the trial type mining rule or the indicator type mining rule. The previous clinical trial set of the first mining rule includes the trial data of each clinical trial, and the last clinical trial set includes the target clinical trial and its trial data.
6. The clinical data mining method according to any one of claims 1 to 4, It is characterized in that The target clinical trial and its test data are obtained, and then further include: The test data of each target clinical trial is standardized, and the standardization includes standardization of at least one of indication information, drug information, sponsor information, test date, test area, number of recruits, test plan, and test status.
7. A clinical data mining device, It is characterized in that include: A collection unit, used to collect clinical trial texts; An extraction unit, used to extract the test data of each clinical trial from the clinical trial text; A screening unit, for screening the test data of each clinical trial based on a preset mining rule to obtain a target clinical trial and its test data; the preset mining rule is determined based on at least one of a clinical trial stage, a threshold of the number of people recruited for an indication, a clinical trial type, and an endpoint indicator type; the clinical trial type and the endpoint indicator type are obtained by entity recognition or rule-based matching for the test scheme in the test data; The target clinical trials include core clinical trials, which include registration clinical trials initiated by developers and / or high-quality randomized controlled clinical trials initiated by researchers; The preset mining rules include at least one of the following rules: Trial stage mining rules are used to mine clinical trials registered and submitted under the preset clinical trial stage; The recruitment number mining rule is used to mine clinical trials with recruitment numbers greater than the corresponding indication recruitment number threshold; Trial type mining rules, used to mine clinical trials under the target trial type; The indicator type mining rule is used to mine clinical trials under the target indicator type, and the target indicator type includes the main endpoint indicator related to effectiveness.
8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, It is characterized in that When the processor executes the program, the steps of the clinical data mining method according to any one of claims 1 to 6 are implemented.
9. A non-transitory computer-readable storage medium having a computer program stored thereon, It is characterized in that When the computer program is executed by a processor, the steps of the clinical data mining method according to any one of claims 1 to 6 are implemented.
