A method, system, device, and medium for enrollment identification of a clinical trial subject

Abstract

This invention provides a method, system, device, and medium for identifying subjects in clinical trials. The method includes: in response to an enrollment identification command for a target patient, obtaining the patient identifier of the target patient; acquiring multiple subject data corresponding to the target patient in parallel from multiple business systems through multiple preset data acquisition channels; wherein the multiple subject data belong to different data modality categories; determining the clinical trial to which the target patient is to be enrolled and obtaining the corresponding set of inclusion and exclusion conditions; mapping each inclusion and exclusion condition in the set of inclusion and exclusion conditions to its corresponding data modality category, obtaining the correspondence between each inclusion and exclusion condition and the data modality category; for each inclusion and exclusion condition, using an inference model under the corresponding data modality category to identify the subject data for enrollment, obtaining the enrollment identification result of the target patient. This invention reduces the amount of operation caused by switching between multiple systems and improves the efficiency of enrollment identification.

Description

Technical Field

[0001] This invention application relates to the field of data identification, and more particularly to a method, system, device, and medium for identifying subjects enrolled in clinical trials. Background Technology

[0002] The identification and enrollment of clinical trial participants is a crucial step in the clinical trial process. During a clinical trial, participants typically need to go through a series of procedures, including signing informed consent forms, entering a screening period, and completing relevant examinations and tests. Then, based on the inclusion and exclusion criteria (hereinafter referred to as inclusion and exclusion criteria) set in the clinical trial protocol, combined with the participant's medical record data and examination and test data, it is necessary to determine whether the participant meets the inclusion criteria for the clinical trial, i.e., to identify the participant for enrollment.

[0003] In existing enrollment identification schemes, reviewers typically search for corresponding subject data in various medical institutions' operational systems for each inclusion / exclusion criterion in the clinical trial protocol. They then manually compare the found data against the criteria for that inclusion / exclusion criterion to determine the enrollment result. However, clinical trial inclusion / exclusion criteria often encompass multiple aspects, including basic subject information, disease characteristics, past medical history, and various examination and testing indicators. Furthermore, the subject data involved in each inclusion / exclusion criterion is usually scattered across different medical operational systems. Existing schemes require reviewers to frequently switch between multiple operational systems during the identification process, and they must manually establish and record the correspondence between each inclusion / exclusion criterion and the required data. When there are many inclusion / exclusion criteria or many operational systems involved, the identification efficiency becomes low, and data mismatches or omissions are easily caused by manual analysis errors or switching between multiple systems, thus affecting the accuracy of the enrollment identification results. Summary of the Invention

[0004] This invention application provides a method, system, device, and medium for identifying subjects in clinical trials, in order to solve the technical problem of how to improve the efficiency of subject identification.

[0005] To address the aforementioned technical problems, this invention provides a method for identifying and enrolling clinical trial subjects, comprising: In response to an enrollment identification instruction for a target patient, the patient identifier of the target patient is obtained; Based on the patient identifier, multiple subject data corresponding to the target patient are acquired in parallel from multiple business systems through multiple preset data acquisition channels; wherein, the multiple subject data belong to different data modal categories; Based on the aforementioned data from various subjects, the clinical trials to be enrolled in by the target patient are determined, and the set of inclusion and exclusion criteria corresponding to the clinical trials is obtained. Map each inbound / outbound condition in the set of inbound / outbound conditions to its corresponding data modality category to obtain the correspondence between each inbound / outbound condition and the data modality category; Based on the aforementioned correspondence, for each of the aforementioned inclusion and exclusion conditions, an inference model under the corresponding data modality category is used to identify the subjects for inclusion, thereby obtaining the inclusion identification results of the target patients for the clinical trial.

[0006] As a preferred embodiment, the data modality category includes at least two of the following: text, image, document, and medical image. Therefore, for each of the aforementioned inclusion and exclusion conditions, an inference model under the corresponding data modality category is used to identify the subjects for enrollment, resulting in the enrollment identification results of the target patients for the clinical trial, including: If the data modality category corresponding to the inclusion / exclusion condition is the text class, the subject data corresponding to the text class is input into the preset inference model; If the data modality category corresponding to the inclusion / exclusion condition is the image class or document class, perform text recognition on the subject data corresponding to the image class or document class to obtain recognized text data, and input the recognized text data into the inference model; If the data modality category corresponding to the ingress / exgress condition is the medical imaging category, image feature extraction is performed on the subject data corresponding to the medical imaging category to obtain image feature data, and the image feature data is input into the inference model; Based on the output of the reasoning model, the enrollment identification results of the target patients for the clinical trial are obtained.

[0007] As a preferred embodiment, based on the patient identifier, the method of acquiring various subject data corresponding to the target patient in parallel from multiple business systems through multiple preset data acquisition channels includes: Based on the patient identifier, the subject data of the text class is obtained from the electronic medical record system through the data acquisition channel corresponding to the text class; Based on the patient identifier, the subject data of the image class is obtained from the medical record archiving system through the data acquisition channel corresponding to the image class; Based on the patient identifier, the subject data of the document class is obtained from the medical record archiving system through the data acquisition channel corresponding to the document class; Based on the patient identifier, the subject data of the medical image category is obtained from the image archiving and communication system through the data acquisition channel corresponding to the medical image category; Based on the patient identifier, the subject data of the structured data class is obtained from the hospital information system through the data acquisition channel corresponding to the structured data class.

[0008] As a preferred embodiment, each inclusion / exclusion condition in the inclusion / exclusion condition set corresponds to the original sequence number in the protocol of the clinical trial, and the correspondence includes the original sequence number; Based on the correspondence, for each inclusion / exclusion condition, an inference model under the corresponding data modality category is used to identify the subject data for enrollment, obtaining the enrollment identification results of the target patient for the clinical trial, including: For each of the aforementioned inclusion and exclusion conditions, an inference model under the corresponding data modality category is used to identify the subjects' data for inclusion, thereby obtaining the sub-identification results of the target patient for that inclusion and exclusion condition; Based on the correspondence, the sub-identification result is bound to the original sequence number of the entry / exit condition; The sub-identification results are sorted in ascending order according to the original serial number to obtain the enrollment identification results of the target patient for the clinical trial.

[0009] As a preferred embodiment, mapping each inbound / outbound condition in the inbound / outbound condition set to its corresponding data modality category to obtain the correspondence between each inbound / outbound condition and the data modality category includes: For each of the in and out conditions in the set of in and out conditions, the matching degree between it and each of the data modality categories is calculated using a preset data-category similarity function; Each of the in / out conditions is mapped to the data modality category with the highest matching degree to obtain the correspondence between each in / out condition and the data modality category.

[0010] As a preferred embodiment, the various subject data include structured subject data; The process of determining the clinical trial to be enrolled in for the target patient based on the various subject data, and obtaining the set of inclusion and exclusion criteria corresponding to the clinical trial, includes: Extract the trial identifiers from the structured subject data; Based on the proposed enrollment trial identifier, the corresponding clinical trial is identified as the proposed enrollment clinical trial for the target patient, and the corresponding set of inclusion and exclusion criteria is obtained from a preset clinical trial database.

[0011] As a preferred embodiment, the step of obtaining the patient identifier of the target patient in response to an enrollment identification instruction for the target patient includes: The system receives the enrollment identification instruction input from an external terminal; wherein the enrollment identification instruction is generated by the external terminal in response to a trigger operation on a preset function control, and the enrollment identification instruction carries the patient identifier of the target patient; Obtain the patient identifier of the target patient from the enrollment identification instruction.

[0012] Accordingly, this invention application also provides an enrollment identification system for clinical trial subjects, including an identifier acquisition module, a data acquisition module, a set acquisition module, a mapping module, and an identification module; wherein, The identifier acquisition module is used to acquire the patient identifier of the target patient in response to the enrollment identification instruction for the target patient; The data acquisition module is used to acquire, based on the patient identifier, multiple subject data corresponding to the target patient from multiple business systems in parallel through multiple preset data acquisition channels; wherein, the multiple subject data belong to different data modality categories; The set acquisition module is used to determine the clinical trial to be enrolled in by the target patient based on the various subject data, and to acquire the set of inclusion and exclusion criteria corresponding to the clinical trial. The mapping module is used to map each inbound / outbound condition in the inbound / outbound condition set to its corresponding data modality category, thereby obtaining the correspondence between each inbound / outbound condition and the data modality category. The identification module is used to identify the subject data for each inclusion / exclusion condition based on the correspondence relationship, using an inference model under the corresponding data modality category, to obtain the inclusion identification result of the target patient for the clinical trial.

[0013] As a preferred embodiment, the data modality category includes at least two of the following: text, image, document, and medical image. For each inclusion / exclusion condition, the identification module uses an inference model under the corresponding data modality category to identify the subject data for inclusion, obtaining the inclusion identification results of the target patient for the clinical trial, including: If the data modality category corresponding to the inclusion / exclusion condition is the text class, the recognition module will input the subject data corresponding to the text class into the preset inference model; If the data modality category corresponding to the inclusion / exclusion condition is the image class or document class, the recognition module performs text recognition on the subject data corresponding to the image class or document class to obtain the recognized text data, and inputs the recognized text data into the inference model; If the data modality category corresponding to the ingress / exgress condition is the medical imaging category, image feature extraction is performed on the subject data corresponding to the medical imaging category to obtain image feature data, and the image feature data is input into the inference model; Based on the output of the reasoning model, the enrollment identification results of the target patients for the clinical trial are obtained.

[0014] As a preferred embodiment, the data acquisition module, based on the patient identifier, acquires various subject data corresponding to the target patient in parallel from multiple business systems through multiple preset data acquisition channels, including: The data acquisition module acquires the subject data of the text class from the electronic medical record system based on the patient identifier and through the data acquisition channel corresponding to the text class; Based on the patient identifier, the subject data of the image class is obtained from the medical record archiving system through the data acquisition channel corresponding to the image class; Based on the patient identifier, the subject data of the document class is obtained from the medical record archiving system through the data acquisition channel corresponding to the document class; Based on the patient identifier, the subject data of the medical image category is obtained from the image archiving and communication system through the data acquisition channel corresponding to the medical image category; Based on the patient identifier, the subject data of the structured data class is obtained from the hospital information system through the data acquisition channel corresponding to the structured data class.

[0015] As a preferred embodiment, each inclusion / exclusion condition in the inclusion / exclusion condition set corresponds to the original sequence number in the protocol of the clinical trial, and the correspondence includes the original sequence number; Based on the correspondence, the identification module uses an inference model under the corresponding data modality category to identify the subject data for each inclusion / exclusion condition, thereby obtaining the inclusion identification results of the target patient for the clinical trial, including: For each inclusion / exclusion condition, the identification module uses an inference model under the corresponding data modality category to identify the subject data for inclusion, and obtains the sub-identification result of the target patient for that inclusion / exclusion condition. Based on the correspondence, the sub-identification result is bound to the original sequence number of the entry / exit condition; The sub-identification results are sorted in ascending order according to the original serial number to obtain the enrollment identification results of the target patient for the clinical trial.

[0016] As a preferred embodiment, the mapping module maps each inbound / outbound condition in the inbound / outbound condition set to its corresponding data modality category, thereby obtaining the correspondence between each inbound / outbound condition and the data modality category, including: For each of the in and out conditions in the set of in and out conditions, the mapping module calculates the matching degree between the in and out condition and each of the data modality categories using a preset data-category similarity function. Each of the in / out conditions is mapped to the data modality category with the highest matching degree to obtain the correspondence between each in / out condition and the data modality category.

[0017] As a preferred embodiment, the various subject data include structured subject data; The set acquisition module, based on the various subject data, determines the clinical trial to which the target patient is to be enrolled, and acquires the set of inclusion and exclusion criteria corresponding to the clinical trial, including: The set acquisition module extracts the proposed trial identifier from the structured subject data; Based on the proposed enrollment trial identifier, the corresponding clinical trial is identified as the proposed enrollment clinical trial for the target patient, and the corresponding set of inclusion and exclusion criteria is obtained from a preset clinical trial database.

[0018] As a preferred embodiment, the identifier acquisition module, in response to an enrollment identification command for a target patient, acquires the patient identifier of the target patient, including: The identification acquisition module receives the enrollment identification instruction input from an external terminal; wherein, the enrollment identification instruction is generated by the external terminal in response to a trigger operation on a preset function control, and the enrollment identification instruction carries the patient identification of the target patient; Obtain the patient identifier of the target patient from the enrollment identification instruction.

[0019] Accordingly, this application also provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the clinical trial subject enrollment identification method.

[0020] Accordingly, this application also provides a computer-readable storage medium, the computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the clinical trial subject enrollment identification method.

[0021] Compared with the prior art, this invention application has the following beneficial effects: This invention application provides a method, system, device, and medium for identifying subjects enrolled in clinical trials. The method includes: in response to an enrollment identification instruction for a target patient, obtaining a patient identifier for the target patient; based on the patient identifier, acquiring multiple subject data corresponding to the target patient in parallel from multiple business systems through multiple preset data acquisition channels; wherein the multiple subject data belong to different data modality categories; based on the multiple subject data, determining the clinical trial to which the target patient is to be enrolled, and obtaining a set of inclusion and exclusion conditions corresponding to the clinical trial; mapping each inclusion and exclusion condition in the set of inclusion and exclusion conditions to its corresponding data modality category, obtaining a correspondence between each inclusion and exclusion condition and the data modality category; based on the correspondence, for each inclusion and exclusion condition, using an inference model under the corresponding data modality category to perform enrollment identification on the subject data, obtaining the enrollment identification result of the target patient for the clinical trial. This invention application obtains a patient identifier in response to an enrollment identification command for a target patient, and acquires multiple subject data in parallel through multiple preset data acquisition channels based on the patient identifier. This allows reviewers to obtain multiple subject data required for identification by triggering an enrollment identification command only once, eliminating the need to repeatedly switch between multiple business systems to search for data as in existing solutions. This reduces the amount of manual operation caused by switching between multiple systems and improves the efficiency of enrollment identification. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating an embodiment of the clinical trial subject enrollment and identification method provided in this invention application.

[0023] Figure 2 This is a schematic diagram of the software interface for one embodiment of the clinical trial subject enrollment identification method provided in this invention application.

[0024] Figure 3 A schematic diagram of the software interface for another embodiment of the clinical trial subject enrollment identification method provided in this invention application.

[0025] Figure 4 This is a schematic diagram of an embodiment of the clinical trial subject enrollment identification system provided in this invention application. Detailed Implementation

[0026] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] Example 1 According to relevant technical records, existing enrollment identification schemes typically involve reviewers searching for corresponding subject data in various medical institutions' operational systems for each inclusion / exclusion criterion in the clinical trial protocol. The data is then manually compared against the criteria for that inclusion / exclusion criterion to determine the enrollment result. However, clinical trial inclusion / exclusion criteria often encompass multiple aspects, including basic subject information, disease characteristics, past medical history, and various examination and testing indicators. Furthermore, the subject data involved in each inclusion / exclusion criterion is usually scattered across different medical operational systems. Existing schemes require reviewers to frequently switch between multiple operational systems during the identification process, and they must manually establish and record the correspondence between each inclusion / exclusion criterion and the required data. When there are many inclusion / exclusion criteria or many operational systems involved, the identification efficiency becomes low, and data mismatches or omissions can easily occur due to human error or switching between multiple systems, thus affecting the accuracy of the enrollment identification results. Secondly, the timeliness of the review is difficult to guarantee. The dynamic changes in core physiological indicators such as blood counts, liver and kidney function, and electrocardiograms mean that manual data acquisition and updates are significantly delayed, which cannot meet the real-time review requirements of clinical trials and is prone to misjudgment. In addition, the lack of uniformity in review standards and the differences in professional level and clinical experience among different reviewers, as well as their different understandings of the ambiguous inclusion and exclusion conditions, all affect the accuracy of the enrollment identification results.

[0028] For one or more of the above technical issues, please refer to Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the clinical trial subject enrollment and identification method provided in this invention application.

[0029] Figure 1 The illustrated embodiment includes steps S101 to S105; each step is described in detail below: Step S101: In response to the enrollment identification instruction for the target patient, obtain the patient identifier of the target patient.

[0030] Specifically, in the enrollment identification scenario of a clinical trial, the target patient is a patient who has signed an informed consent form and entered the screening period, awaiting assessment of whether they meet the clinical trial enrollment criteria; the patient identifier is unique identification information within the medical institution that can uniquely identify the target patient, such as a unique patient identification number assigned to the target patient by the medical institution. The enrollment identification command is an instruction used to trigger the enrollment identification process for the target patient. This step obtains the target patient's patient identifier in response to the enrollment identification command, serving as an index for subsequently retrieving various subject data corresponding to the target patient from multiple business systems in parallel.

[0031] As a preferred embodiment, the step of obtaining the patient identifier of the target patient in response to an enrollment identification instruction for the target patient includes: The system receives the enrollment identification instruction input from an external terminal; wherein the enrollment identification instruction is generated by the external terminal in response to a trigger operation on a preset function control, and the enrollment identification instruction carries the patient identifier of the target patient; Obtain the patient identifier of the target patient from the enrollment identification instruction.

[0032] Specifically, such as Figure 2 As shown, Figure 2 The operation shown can include the target patient in the enrollment identification process. The external terminal can be a terminal device with a doctor's workstation deployed on it, and the preset function control can be a function control deployed in a designated functional area of ​​the doctor's workstation, such as a function button on the doctor's workstation interface used to trigger enrollment identification. The triggering operation can be any one of a click operation, a touch operation, or a voice activation operation on the preset function control. In this embodiment, the reviewer (such as the doctor responsible for enrollment review) performs the triggering operation (such as clicking the function button) on the doctor's workstation for the target patient to be reviewed. In response to the triggering operation, the external terminal encapsulates the patient identifier corresponding to the target patient into the enrollment identification instruction and sends the enrollment identification instruction to the device executing this method. After receiving the enrollment identification instruction, the device executing this method parses the patient identifier from the instruction, thereby completing the acquisition of the patient identifier of the target patient. By associating the triggering of the enrollment identification instruction with the preset function control, the reviewer only needs to perform a triggering operation on the preset function control once to start the enrollment identification process for the target patient, which facilitates seamless triggering of enrollment identification in the doctor's workstation.

[0033] Step S102: Based on the patient identifier, multiple subject data corresponding to the target patient are acquired in parallel from multiple business systems through multiple preset data acquisition channels; wherein the multiple subject data belong to different data modality categories.

[0034] Specifically, the business system refers to an information system deployed within a medical institution that stores data related to the target patient. The medical records, examination and test data generated by the target patient during the clinical trial screening period and earlier during the consultation period are usually distributed across multiple business systems.

[0035] The data acquisition channel refers to the channel between the device executing this method and each of the business systems for real-time data interaction. Each data acquisition channel corresponds to a data modality category and points to at least one of the business systems.

[0036] The specific implementation of the data acquisition channel can be a database query channel based on Structured Query Language (SQL) or a data interaction channel based on Application Programming Interface (API). This embodiment does not limit this to either.

[0037] The subject data refers to data generated by the target patient within the medical institution that can be used for enrollment identification, such as outpatient records, admission records, laboratory reports, examination reports, nursing records, medical images, and basic patient information. The data modality category is a classification of subject data based on its presentation format. Subject data under different data modality categories differ in data format, such as natural language format, image format, document format, or medical image format.

[0038] In this step, the device executing this method uses the patient identifier of the target patient as an index and initiates data query requests in parallel to multiple business systems through multiple preset data acquisition channels. The multiple business systems then return various subject data corresponding to the target patient in parallel. Compared to existing solutions that require switching between multiple business systems one by one to search for data, this step, through multi-channel parallel acquisition, can complete the collection of all required subject data in a single response to the enrollment identification command, thereby significantly reducing the manual burden of cross-system operations and ensuring data real-time performance.

[0039] As a preferred embodiment, based on the patient identifier, the method of acquiring various subject data corresponding to the target patient in parallel from multiple business systems through multiple preset data acquisition channels includes: Based on the patient identifier, the subject data of the text class is obtained from the electronic medical record system through the data acquisition channel corresponding to the text class; Based on the patient identifier, the subject data of the image class is obtained from the medical record archiving system through the data acquisition channel corresponding to the image class; Based on the patient identifier, the subject data of the document class is obtained from the medical record archiving system through the data acquisition channel corresponding to the document class; Based on the patient identifier, the subject data of the medical image category is obtained from the image archiving and communication system through the data acquisition channel corresponding to the medical image category; Based on the patient identifier, the subject data of the structured data class is obtained from the hospital information system through the data acquisition channel corresponding to the structured data class.

[0040] Specifically, the text-based subject data refers to subject data recorded in natural language text form. The electronic medical record system is a business system within a medical institution used to record and store natural language data generated during patient diagnosis and treatment. The device executing this method initiates a request to the electronic medical record system through the data acquisition channel corresponding to the text-based data, using the patient identifier as the query condition, to obtain the text-based subject data corresponding to the target patient. The text-based subject data may include, for example, at least one of the target patient's outpatient records, admission records, laboratory reports, examination reports, and nursing records.

[0041] The image-based subject data refers to subject data archived in image format, and the document-based subject data refers to subject data archived in document format such as Portable Document Format (PDF). The medical record archiving system is a business system within a medical institution used to archive and store medical record-related image and document data. The device executing this method initiates a request to the medical record archiving system through the data acquisition channel corresponding to the image type, using the patient identifier as the query condition, to obtain the image-based subject data corresponding to the target patient; and through the data acquisition channel corresponding to the document type, it initiates a request to the medical record archiving system through the patient identifier as the query condition, to obtain the document-based subject data corresponding to the target patient.

[0042] The medical imaging subject data refers to uninterpreted raw medical imaging data. The imaging archiving and communication system is a business system within a medical institution used to store and transmit medical imaging data. The device executing this method initiates a request to the imaging archiving and communication system through the data acquisition channel corresponding to the medical imaging category, using the patient identifier as the query condition, to obtain the medical imaging subject data corresponding to the target patient. The medical imaging subject data may include, for example, at least one of computed tomography (CT) images and magnetic resonance (MR) images.

[0043] The structured data class subject data refers to subject data recorded with a preset field structure that can be directly parsed by a database. The hospital information system is a business system within a medical institution used to manage basic patient information and structured fields related to the proposed enrollment trial. The device executing this method initiates a request to the hospital information system through the data acquisition channel corresponding to the structured data class, using the patient identifier as the query condition, to obtain the structured data class subject data corresponding to the target patient. The structured data class subject data may include, for example, at least one of the target patient's gender, date of birth, proposed clinical trial code, current trial stage status, informed consent signing time, and proposed enrollment trial identifier.

[0044] Through the above correspondence, the device executing this method can selectively and in parallel retrieve subject data of different modalities from their original business systems, avoiding confusion between subject data of different modalities during the collection process, and facilitating the use of different processing logics for different modalities in subsequent steps.

[0045] Step S103: Based on the various subject data, determine the clinical trial to be enrolled in by the target patient, and obtain the set of inclusion and exclusion criteria corresponding to the clinical trial.

[0046] Specifically, the clinical trial is a medical research project that is pre-registered in the clinical trial database and is available for patients to participate in. Each clinical trial corresponds to a set of inclusion and exclusion criteria for determining whether a subject meets the inclusion criteria. The set of inclusion and exclusion criteria is the set of inclusion and exclusion criteria constituted by that set of inclusion and exclusion criteria.

[0047] In this step, the device executing this method determines the clinical trial the target patient intends to participate in based on the various subject data obtained in step S102, and then obtains the set of inclusion and exclusion criteria corresponding to the determined clinical trial, which serves as the basis for subsequent enrollment identification. The number of inclusion and exclusion criteria included in the set is related to the complexity of the clinical trial. For example, for tumor-related clinical trials, the number of inclusion and exclusion criteria in the set can reach 60 or more, covering multiple aspects such as the patient's basic information, disease characteristics, past medical history, and various examination and testing indicators.

[0048] As a preferred embodiment, the various subject data include structured subject data; The process of determining the clinical trial to be enrolled in for the target patient based on the various subject data, and obtaining the set of inclusion and exclusion criteria corresponding to the clinical trial, includes: Extract the trial identifiers from the structured subject data; Based on the proposed enrollment trial identifier, the corresponding clinical trial is identified as the proposed enrollment clinical trial for the target patient, and the corresponding set of inclusion and exclusion criteria is obtained from a preset clinical trial database.

[0049] Specifically, the structured subject data refers to the subject data of the aforementioned structured data class, which can be obtained from the hospital information system; the proposed enrollment trial identifier is an identifier used to uniquely identify the clinical trial in which the target patient intends to participate, such as a string like "Internal Medicine-2025-026" to uniquely identify the clinical trial in which the target patient intends to participate. In the actual workflow of medical institutions, before the target patient signs the informed consent form for the clinical trial and enters the screening period, the relevant doctor usually registers the proposed enrollment trial identifier corresponding to the clinical trial in which the target patient intends to participate in the clinical trial into the target patient's structured data in the hospital information system, as a marker for the target patient's intended participation in the clinical trial.

[0050] In this step, the device executing this method first extracts the proposed enrollment trial identifier corresponding to the target patient from the structured subject data obtained in step S102. If the proposed enrollment trial identifier cannot be extracted from the structured subject data, it indicates that the target patient has no registration record for participating in a clinical trial, and the enrollment identification process for the target patient can be terminated directly, and the corresponding prompt information can be output to the external terminal. If the proposed enrollment trial identifier is extracted, the clinical trial corresponding to the proposed enrollment trial identifier is determined as the clinical trial for which the target patient is to be enrolled, and the set of inclusion and exclusion conditions corresponding to the clinical trial for which the proposed enrollment is to be enrolled is retrieved from the preset clinical trial database using the proposed enrollment trial identifier as the query condition. The pre-set clinical trial database is a database that has been pre-registered with various clinical trials and their corresponding inclusion and exclusion criteria. The inclusion and exclusion criteria stored in the clinical trial database may include, for example, whether the age meets the requirements when signing the informed consent form, whether it is confirmed by histology or cytology, whether there are measurable lesions that meet the RECIST V1.1 criteria, whether there is mismatch repair deficiency (dMMR) or microsatellite instability-high (MSI-H), whether there is a cardiovascular event of a specified level, and whether the indicators of the blood system such as neutrophil count, platelet count, and hemoglobin meet the requirements.

[0051] In this way, the device executing this method can automatically determine the clinical trial to be enrolled based on the registered enrollment information of the target patient, and automatically obtain the set of inclusion and exclusion criteria corresponding to the clinical trial. This eliminates the need for reviewers to manually input clinical trial information or manually search for inclusion and exclusion criteria, and can promptly exclude patients who have not registered for enrollment trials to avoid invalid enrollment identification.

[0052] Step S104: Map each inbound / outbound condition in the set of inbound / outbound conditions to its corresponding data modality category to obtain the correspondence between each inbound / outbound condition and the data modality category.

[0053] Specifically, the data modalities of the subject data on which different inclusion and exclusion criteria in the set of inclusion and exclusion criteria should be based may be different. For example, the inclusion and exclusion criterion "age ≥ 18 years at the time of signing informed consent" can be determined based on text-based subject data in natural language; the inclusion and exclusion criterion "patients have no of the following cardiovascular diseases as determined by electrocardiogram and color Doppler ultrasound examinations:..." can be determined based on image-based or document-based subject data corresponding to the examination report; and the inclusion and exclusion criterion "at least one measurable lesion meeting RECIST V1.1 criteria" needs to be determined based on raw medical imaging subject data.

[0054] In this step, the device executing this method analyzes the data modality category that should be used to determine each inclusion / exclusion condition in the inclusion / exclusion condition set, and maps the inclusion / exclusion condition to the analyzed data modality category, thereby obtaining the correspondence between each inclusion / exclusion condition in the inclusion / exclusion condition set and its corresponding data modality category. This correspondence is used to indicate the data modality category of the subject data to be used in subsequent steps when identifying inclusion for each inclusion / exclusion condition.

[0055] As a preferred embodiment, mapping each inbound / outbound condition in the inbound / outbound condition set to its corresponding data modality category to obtain the correspondence between each inbound / outbound condition and the data modality category includes: For each of the in and out conditions in the set of in and out conditions, the matching degree between it and each of the data modality categories is calculated using a preset data-category similarity function; Each of the in / out conditions is mapped to the data modality category with the highest matching degree to obtain the correspondence between each in / out condition and the data modality category.

[0056] Specifically, the data-category similarity function is a pre-configured function used to measure the similarity between any in / out condition and any data modality category. The data-category similarity function takes the in / out condition and the data modality category as input and outputs a matching degree. The matching degree indicates the suitability of the in / out condition when judged using subject data within that data modality category; a higher matching degree indicates that the subject data within that data modality category is more suitable for judging the in / out condition. The data-category similarity function can be implemented by a pre-trained similarity discrimination model or by preset rules based on keywords or semantic features; this embodiment does not limit this.

[0057] In this step, the device executing this method inputs each ingress / exgress condition in the ingress / exgress condition set, along with each data modality category, into the data-category similarity function to obtain the matching degree between the ingress / exgress condition and each data modality category. Then, it compares the multiple matching degrees corresponding to each ingress / exgress condition and maps it to the data modality category with the highest matching degree. In the data-category similarity function, the weight distribution can be controlled by hyperparameters to adapt to the characteristics of different clinical trials. For example, for clinical trials where text-based ingress / exgress conditions are predominant, the weight of text-based similarity in the data-category similarity function can be increased to make the matching degree between text-based ingress / exgress conditions higher than that of other categories. After each ingress / exgress condition in the ingress / exgress condition set has been mapped as described above, the correspondence between each ingress / exgress condition in the ingress / exgress condition set and the data modality category is obtained.

[0058] Step S105: Based on the correspondence, for each of the inclusion and exclusion conditions, the subject data is identified by the inference model under the corresponding data modality category to obtain the inclusion identification result of the target patient for the clinical trial.

[0059] Specifically, in this step, the device executing this method determines the data modality category corresponding to each inclusion / exclusion condition in the inclusion / exclusion condition set according to the correspondence obtained in step S104. Then, it selects subject data under the determined data modality category from the multiple subject data obtained in step S102, and performs enrollment identification based on the selected subject data for that inclusion / exclusion condition to obtain the enrollment identification result under that condition. After enrollment identification has been completed for all inclusion / exclusion conditions in the inclusion / exclusion condition set, the enrollment identification results corresponding to each inclusion / exclusion condition are integrated to obtain the enrollment identification result of the target patient for the clinical trial.

[0060] As a preferred embodiment, the data modality category includes at least two of the following: text, image, document, and medical image. Therefore, for each of the aforementioned inclusion and exclusion conditions, an inference model under the corresponding data modality category is used to identify the subjects for enrollment, resulting in the enrollment identification results of the target patients for the clinical trial, including: If the data modality category corresponding to the inclusion / exclusion condition is the text class, the subject data corresponding to the text class is input into the preset inference model; If the data modality category corresponding to the inclusion / exclusion condition is the image class or document class, perform text recognition on the subject data corresponding to the image class or document class to obtain recognized text data, and input the recognized text data into the inference model; If the data modality category corresponding to the ingress / exgress condition is the medical imaging category, image feature extraction is performed on the subject data corresponding to the medical imaging category to obtain image feature data, and the image feature data is input into the inference model; Based on the output of the reasoning model, the enrollment identification results of the target patients for the clinical trial are obtained.

[0061] Specifically, the inference model is a preset model that takes the in and out conditions and their corresponding textual judgment input data as input and the group identification result as output. The inference model can be implemented by a large language model. For example, it can be a model based on a general large language model and adapted to the task through at least one of the following techniques: Prompt Engineer, Content Engineer, Retrieval-Augmented Generation (RAG), Chain of Thought (CoT), and Cache-Augmented Generation (CAG), or a model obtained after pre-training and fine-tuning. This embodiment does not limit the specific implementation form of the inference model.

[0062] For the inclusion / exclusion condition corresponding to the text class data modality, since the subject data of the text class is already in natural language form, the inclusion / exclusion condition and the corresponding subject data of the text class can be directly input into the inference model. The inference model will then infer the content related to the inclusion / exclusion condition in the subject data of the text class and output the inclusion / exclusion result under the inclusion / exclusion condition.

[0063] For inclusion / exclusion conditions corresponding to the image or document modality categories, since the content carried in the subject data of the image or document categories is not in natural language form that can be directly processed by the inference model, text recognition is first performed on the subject data corresponding to the image or document categories to obtain recognized text data. The text recognition can be implemented by an Optical Character Recognition (OCR) algorithm or a visual model, and a preset confidence threshold can be set. Only recognition results with a confidence level higher than the confidence threshold are retained as the recognized text data to improve recognition accuracy. Then, the inclusion / exclusion condition and its corresponding recognized text data are input into the inference model, which performs inference based on the recognized text data and outputs the inclusion / exclusion recognition result under the inclusion / exclusion condition.

[0064] For the inclusion / exclusion criteria corresponding to the medical imaging category, since the subject data of the medical imaging category is raw image data, information such as lesion location and size cannot be directly obtained from the raw images in natural language form. Therefore, image feature extraction is first performed on the subject data corresponding to the medical imaging category to obtain image feature data. The image feature extraction can be implemented by an image segmentation model. The image feature data may include, for example, image features quantified in text form such as lesion location and lesion size. Then, the inclusion / exclusion criteria and the corresponding image feature data are input into the inference model, and the inference model performs inference based on the image feature data and outputs the inclusion / exclusion criteria.

[0065] In this way, subject data from different data modalities can be preprocessed and converted into an input format that the inference model can process, thereby enabling the use of the same inference model to identify inclusion and exclusion conditions for different data modalities.

[0066] As a preferred embodiment, each inclusion / exclusion condition in the inclusion / exclusion condition set corresponds to the original sequence number in the protocol of the clinical trial, and the correspondence includes the original sequence number; Based on the correspondence, for each inclusion / exclusion condition, an inference model under the corresponding data modality category is used to identify the subject data for enrollment, obtaining the enrollment identification results of the target patient for the clinical trial, including: For each of the aforementioned inclusion and exclusion conditions, an inference model under the corresponding data modality category is used to identify the subject data for inclusion, thereby obtaining the sub-identification result of the target patient for that inclusion and exclusion condition; Based on the correspondence, the sub-identification result is bound to the original sequence number of the entry / exit condition; The sub-identification results are sorted in ascending order according to the original serial number to obtain the enrollment identification results of the target patient for the clinical trial.

[0067] Specifically, the clinical trial protocol pre-assigns an original sequence number to each inclusion / exclusion condition in the inclusion / exclusion condition set. This original sequence number indicates the numbering order of the inclusion / exclusion condition within the clinical trial protocol. When the correspondence is obtained in step S104, the original sequence number corresponding to each inclusion / exclusion condition is recorded in the correspondence so that each inclusion / exclusion condition can be identified according to its original sequence number when the sub-identification results are subsequently integrated.

[0068] In this step, the device executing this method, for each inclusion / exclusion condition in the inclusion / exclusion condition set, obtains a sub-identification result for the target patient for that inclusion / exclusion condition according to the aforementioned processing logic. The sub-identification result may include a judgment conclusion regarding the inclusion / exclusion condition and the corresponding reason or basis for that judgment conclusion. Then, based on the original sequence number recorded in the correspondence, the sub-identification result is bound to the original sequence number of the inclusion / exclusion condition. After all inclusion / exclusion conditions in the inclusion / exclusion condition set have undergone the above processing, the sub-identification results are sorted in ascending order of the original sequence numbers corresponding to each inclusion / exclusion condition to obtain the target patient's enrollment identification result for the clinical trial. The target patient's enrollment identification result for the clinical trial can be output to the external terminal, so that each sub-identification result is presented to the reviewer in a manner consistent with the numbering order of the inclusion / exclusion conditions in the clinical trial protocol. This facilitates the reviewer's further review of the enrollment identification result or the writing of a review report using the numbering order of the clinical trial protocol.

[0069] In some implementations, when integrating and outputting the results of each sub-identification, the first sorting dimension can be sorted according to the original sequence number of the in and out conditions, and the second sorting dimension can be optimized according to the hyperparameter control. For example, when the hyperparameter is 1, the output is "judgment result + brief reason", and when the hyperparameter is 2, the output is "judgment result + detailed basis + data source + reason".

[0070] Expected output set It can satisfy: Where y represents the sub-identification result. For the screening results of compliant patients, The classification results are numbered. These are the numbered judgment conclusions. The recognition results are in a standardized integrated format sorted by serial number.

[0071] The integrated judgment results (which can be arranged in order of condition number 1, 2, 3...) are output to the doctor's workstation, supporting "one-click insertion" of medical documents. Doctors can directly use this sequence number to write the review report without manual adjustment, thus completing the review process.

[0072] One type of recognition result is as follows Figure 3 As shown, another example of the generated results is as follows (and...). Figure 3 (The content differs; this is only an example of the text version.) In some preferred embodiments, hyperparameters can be optimized based on human experience, such as by adjusting them through a loss function.

[0073] For example, the loss function can consider classification loss (quantifying the accuracy of numbered classification), discrimination loss (e.g., quantifying the bias of numbered discrimination conclusions by cross-entropy), and integration loss (e.g., quantifying the numbering and the deviation of the format from the standard by edit distance).

[0074] Furthermore, the hyperparameter update formula is as follows: ; Among them, For the first Wheel hyperparameters, For empirically tuned functions, typical rules include: like ,but (Lower the test number matching threshold and relax the screening); like ,but (Lowering the text discrimination threshold improves recall rate); like And if the deviation stems from the format, then (Switch format mode to optimize presentation); If the deviation originates from the sequence number sorting, the system will be triggered to re-verify and bind the in-row condition sequence numbers.

[0075] in, , and The preset threshold, This indicates the results of the compliant patient screening regarding the first... The loss function of the round hyperparameters, This indicates the numbered judgment conclusion regarding the first... The loss function of the round hyperparameters, The recognition results for the standardized integrated format sorted by serial number are related to the first... The loss function of the round hyperparameters, This refers to the hyperparameter adjustment amount.

[0076] Accordingly, such as Figure 4 As shown, this invention application also provides a clinical trial subject enrollment identification system 400, including an identifier acquisition module 401, a data acquisition module 402, a set acquisition module 403, a mapping module 404, and an identification module 405; wherein, The identifier acquisition module 401 is used to acquire the patient identifier of the target patient in response to the enrollment identification instruction for the target patient; The data acquisition module 402 is used to acquire multiple subject data corresponding to the target patient in parallel from multiple business systems through multiple preset data acquisition channels based on the patient identifier; wherein the multiple subject data belong to different data modality categories; The set acquisition module 403 is used to determine the clinical trial to be enrolled in by the target patient based on the multiple subject data, and to acquire the set of inclusion and exclusion criteria corresponding to the clinical trial. The mapping module 404 is used to map each inbound and outbound condition in the inbound and outbound condition set to the corresponding data modality category, thereby obtaining the correspondence between each inbound and outbound condition and the data modality category; The identification module 405 is used to identify the subject data based on the correspondence and for each of the inclusion and exclusion conditions, using an inference model under the corresponding data modality category, to obtain the inclusion identification result of the target patient for the clinical trial.

[0077] As a preferred embodiment, the data modality category includes at least two of the following: text, image, document, and medical image. For each inclusion / exclusion condition, the identification module 405 uses an inference model under the corresponding data modality category to identify the subject data for inclusion, obtaining the inclusion identification result of the target patient for the clinical trial, including: If the data modality category corresponding to the ingress / exgress condition is the text class, the recognition module 405 inputs the subject data corresponding to the text class into the preset inference model; If the data modality category corresponding to the inclusion / exclusion condition is the image class or document class, the recognition module 405 performs text recognition on the subject data corresponding to the image class or document class to obtain the recognized text data, and inputs the recognized text data into the inference model; If the data modality category corresponding to the ingress / exgress condition is the medical imaging category, the identification module 405 extracts image features from the subject data corresponding to the medical imaging category to obtain image feature data, and inputs the image feature data into the inference model. Based on the output of the reasoning model, the enrollment identification results of the target patients for the clinical trial are obtained.

[0078] As a preferred embodiment, the data acquisition module 402, based on the patient identifier, acquires various subject data corresponding to the target patient in parallel from multiple business systems through multiple preset data acquisition channels, including: The data acquisition module 402 acquires the subject data of the text class from the electronic medical record system based on the patient identifier through the data acquisition channel corresponding to the text class; Based on the patient identifier, the subject data of the image class is obtained from the medical record archiving system through the data acquisition channel corresponding to the image class; Based on the patient identifier, the subject data of the document class is obtained from the medical record archiving system through the data acquisition channel corresponding to the document class; Based on the patient identifier, the subject data of the medical image category is obtained from the image archiving and communication system through the data acquisition channel corresponding to the medical image category; Based on the patient identifier, the subject data of the structured data class is obtained from the hospital information system through the data acquisition channel corresponding to the structured data class.

[0079] As a preferred embodiment, each inclusion / exclusion condition in the inclusion / exclusion condition set corresponds to the original sequence number in the protocol of the clinical trial, and the correspondence includes the original sequence number; Based on the correspondence, the identification module 405 uses an inference model under the corresponding data modality category to identify the subject data for each inclusion / exclusion condition, thereby obtaining the inclusion identification result of the target patient for the clinical trial, including: For each inclusion / exclusion condition, the identification module 405 uses an inference model under the corresponding data modality category to identify the subject data for inclusion, and obtains the sub-identification result of the target patient for that inclusion / exclusion condition. Based on the correspondence, the sub-identification result is bound to the original sequence number of the entry / exit condition; The sub-identification results are sorted in ascending order according to the original serial number to obtain the enrollment identification results of the target patient for the clinical trial.

[0080] As a preferred embodiment, the mapping module 404 maps each inbound / outbound condition in the inbound / outbound condition set to its corresponding data modality category, thereby obtaining the correspondence between each inbound / outbound condition and the data modality category, including: For each of the in and out conditions in the set of in and out conditions, the mapping module 404 calculates the matching degree between the in and out conditions and each of the data modality categories using a preset data-category similarity function. Each of the in / out conditions is mapped to the data modality category with the highest matching degree to obtain the correspondence between each in / out condition and the data modality category.

[0081] As a preferred embodiment, the various subject data include structured subject data; The set acquisition module 403, based on the various subject data, determines the clinical trial to which the target patient is to be enrolled, and acquires the set of inclusion and exclusion criteria corresponding to the clinical trial, including: The set acquisition module 403 extracts the proposed trial identifier from the structured subject data; Based on the proposed enrollment trial identifier, the corresponding clinical trial is identified as the proposed enrollment clinical trial for the target patient, and the corresponding set of inclusion and exclusion criteria is obtained from a preset clinical trial database.

[0082] As a preferred embodiment, the identifier acquisition module 401, in response to an enrollment identification command for a target patient, acquires the patient identifier of the target patient, including: The identification acquisition module 401 receives the enrollment identification instruction input from an external terminal; wherein, the enrollment identification instruction is generated by the external terminal in response to a trigger operation on a preset function control, and the enrollment identification instruction carries the patient identification of the target patient; Obtain the patient identifier of the target patient from the enrollment identification instruction.

[0083] Accordingly, this application also provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the clinical trial subject enrollment identification method.

[0084] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the terminal, connecting various parts of the terminal via various interfaces and lines.

[0085] The memory can be used to store the computer program. The processor implements various functions of the terminal by running or executing the computer program stored in the memory and calling the data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0086] Accordingly, this application also provides a computer-readable storage medium, the computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the clinical trial subject enrollment identification method.

[0087] Wherein, if the modules of the device / terminal equipment / system integration are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by a processor, it can implement the steps of the various method embodiments described above. Wherein, the computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0088] Compared with the prior art, this invention application has the following beneficial effects: This invention application provides a method, system, device, and medium for identifying subjects enrolled in clinical trials. The method includes: in response to an enrollment identification instruction for a target patient, obtaining a patient identifier for the target patient; based on the patient identifier, acquiring multiple subject data corresponding to the target patient in parallel from multiple business systems through multiple preset data acquisition channels; wherein the multiple subject data belong to different data modality categories; based on the multiple subject data, determining the clinical trial to which the target patient is to be enrolled, and obtaining a set of inclusion and exclusion conditions corresponding to the clinical trial; mapping each inclusion and exclusion condition in the set of inclusion and exclusion conditions to its corresponding data modality category, obtaining a correspondence between each inclusion and exclusion condition and the data modality category; based on the correspondence, for each inclusion and exclusion condition, using an inference model under the corresponding data modality category to perform enrollment identification on the subject data, obtaining the enrollment identification result of the target patient for the clinical trial. This invention obtains a patient identifier in response to an enrollment identification command for a target patient, and then acquires various subject data in parallel through multiple preset data acquisition channels based on this patient identifier. This allows reviewers to obtain the required subject data for identification with only a single enrollment identification command, eliminating the need for repeated switching between multiple business systems to read medical records and search for data, as in existing solutions. Furthermore, it directly identifies the most relevant evidence and generates a review report based on each condition, thereby reducing the amount of manual work involved in switching between multiple systems, reviewing medical records, and writing reports, and improving the efficiency of enrollment identification. Clinical trials have verified that this invention's technical solution achieved an overall accuracy rate of 98.8% in a review test involving 24 patients and 1080 inclusion / exclusion criteria. The average time for identifying a single patient was reduced from 55.56 minutes to 21.36 minutes, significantly improving the efficiency and standardization of clinical trial enrollment identification and addressing various technical pain points of traditional manual review.

[0089] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for identifying subjects for inclusion in a clinical trial, characterized in that, include: In response to an enrollment identification instruction for a target patient, the patient identifier of the target patient is obtained; Based on the patient identifier, multiple subject data corresponding to the target patient are acquired in parallel from multiple business systems through multiple preset data acquisition channels; wherein, the multiple subject data belong to different data modal categories; Based on the aforementioned data from various subjects, the clinical trials to be enrolled in by the target patient are determined, and the set of inclusion and exclusion criteria corresponding to the clinical trials is obtained. Map each inbound / outbound condition in the set of inbound / outbound conditions to its corresponding data modality category to obtain the correspondence between each inbound / outbound condition and the data modality category; Based on the aforementioned correspondence, for each of the aforementioned inclusion and exclusion conditions, an inference model under the corresponding data modality category is used to identify the subjects for inclusion, thereby obtaining the inclusion identification results of the target patients for the clinical trial.

2. The method for identifying subjects in a clinical trial as described in claim 1, characterized in that, The data modality categories include at least two of the following: text, image, document, and medical image. Therefore, for each of the aforementioned inclusion and exclusion conditions, an inference model under the corresponding data modality category is used to identify the subjects for enrollment, resulting in the enrollment identification results of the target patients for the clinical trial, including: If the data modality category corresponding to the inclusion / exclusion condition is the text class, the subject data corresponding to the text class is input into the preset inference model; If the data modality category corresponding to the inclusion / exclusion condition is the image class or document class, perform text recognition on the subject data corresponding to the image class or document class to obtain recognized text data, and input the recognized text data into the inference model; If the data modality category corresponding to the ingress / exgress condition is the medical imaging category, image feature extraction is performed on the subject data corresponding to the medical imaging category to obtain image feature data, and the image feature data is input into the inference model; Based on the output of the reasoning model, the enrollment identification results of the target patients for the clinical trial are obtained.

3. The method for identifying subjects in a clinical trial as described in claim 2, characterized in that, Based on the patient identifier, the process involves acquiring various subject data corresponding to the target patient in parallel from multiple business systems through multiple preset data acquisition channels, including: Based on the patient identifier, the subject data of the text class is obtained from the electronic medical record system through the data acquisition channel corresponding to the text class; Based on the patient identifier, the subject data of the image class is obtained from the medical record archiving system through the data acquisition channel corresponding to the image class; Based on the patient identifier, the subject data of the document class is obtained from the medical record archiving system through the data acquisition channel corresponding to the document class; Based on the patient identifier, the subject data of the medical image category is obtained from the image archiving and communication system through the data acquisition channel corresponding to the medical image category; Based on the patient identifier, the subject data of the structured data class is obtained from the hospital information system through the data acquisition channel corresponding to the structured data class.

4. The method for identifying and enrolling clinical trial subjects as described in claim 1, characterized in that, Each inclusion / exclusion condition in the inclusion / exclusion condition set corresponds to the original sequence number in the protocol of the clinical trial, and the correspondence includes the original sequence number; Based on the correspondence, for each inclusion / exclusion condition, an inference model under the corresponding data modality category is used to identify the subject data for enrollment, obtaining the enrollment identification results of the target patient for the clinical trial, including: For each of the aforementioned inclusion and exclusion conditions, an inference model under the corresponding data modality category is used to identify the subject data for inclusion, thereby obtaining the sub-identification result of the target patient for that inclusion and exclusion condition; Based on the correspondence, the sub-identification result is bound to the original sequence number of the entry / exit condition; The sub-identification results are sorted in ascending order according to the original serial number to obtain the enrollment identification results of the target patient for the clinical trial.

5. The method for identifying subjects in a clinical trial as described in claim 1, characterized in that, The step of mapping each inbound / outbound condition in the set of inbound / outbound conditions to its corresponding data modality category, thereby obtaining the correspondence between each inbound / outbound condition and the data modality category, includes: For each of the in and out conditions in the set of in and out conditions, the matching degree between it and each of the data modality categories is calculated using a preset data-category similarity function; Each of the in / out conditions is mapped to the data modality category with the highest matching degree to obtain the correspondence between each in / out condition and the data modality category.

6. The method for identifying subjects in a clinical trial as described in claim 1, characterized in that, The various types of subject data include structured subject data; The process of determining the clinical trial to be enrolled in for the target patient based on the various subject data, and obtaining the set of inclusion and exclusion criteria corresponding to the clinical trial, includes: Extract the trial identifiers from the structured subject data; Based on the proposed enrollment trial identifier, the corresponding clinical trial is identified as the proposed enrollment clinical trial for the target patient, and the corresponding set of inclusion and exclusion criteria is obtained from a preset clinical trial database.

7. The method for identifying subjects in a clinical trial as described in claim 1, characterized in that, The step of obtaining the patient identifier of the target patient in response to an enrollment identification instruction for the target patient includes: The system receives the enrollment identification instruction input from an external terminal; wherein the enrollment identification instruction is generated by the external terminal in response to a trigger operation on a preset function control, and the enrollment identification instruction carries the patient identifier of the target patient; Obtain the patient identifier of the target patient from the enrollment identification instruction.

8. A clinical trial subject enrollment identification system, characterized in that, It includes an identifier acquisition module, a data acquisition module, a set acquisition module, a mapping module, and a recognition module; among which, The identifier acquisition module is used to acquire the patient identifier of the target patient in response to the enrollment identification instruction for the target patient; The data acquisition module is used to acquire, based on the patient identifier, multiple subject data corresponding to the target patient from multiple business systems in parallel through multiple preset data acquisition channels; wherein, the multiple subject data belong to different data modality categories; The set acquisition module is used to determine the clinical trial to be enrolled in by the target patient based on the various subject data, and to acquire the set of inclusion and exclusion criteria corresponding to the clinical trial. The mapping module is used to map each inbound / outbound condition in the inbound / outbound condition set to its corresponding data modality category, thereby obtaining the correspondence between each inbound / outbound condition and the data modality category. The identification module is used to identify the subject data for each inclusion / exclusion condition based on the correspondence relationship, using an inference model under the corresponding data modality category, to obtain the inclusion identification result of the target patient for the clinical trial.

9. A terminal device, characterized in that, It includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the clinical trial subject enrollment identification method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device on which the computer-readable storage medium is located to perform the clinical trial subject enrollment identification method as described in any one of claims 1 to 7.

Images