How to (Semi)-Automatically Spot Prescreening Oriented Eligibility Criteria.

Area of Science:

• Biomedical Informatics
• Clinical Trial Operations
• Health Data Science

Background:

• Clinical Trial Recruitment Support Systems (CTRSS) increasingly use Electronic Health Records (EHR) for patient-trial matching.
• Manual processing of free-text clinical trial eligibility criteria (EC) for EHR querying is time-consuming and inefficient.
• Automating patient-trial matching requires structured eligibility criteria suitable for EHR querying.

Purpose of the Study:

• To develop a methodological approach for semi-automatically detecting Prescreening-Oriented Eligibility Criteria (POEC).
• To build a reusable library of POEC for the development and evaluation of EHR-based CTRSS.
• To facilitate the use of EHR data for improved clinical trial participation.

Main Methods:

• Decomposition of free-text EC into standardized elements.
• Development of a rule-based algorithm for semi-automatic POEC detection.
• Annotation of 381 free-text EC from 20 cancer clinical trials using a framework of 96 elementary EC patterns across 17 domains.
• Creation of a publicly available POEC library (PENELOPE POEC library) fed by the PENELOPE-C2Q pipeline.

Main Results:

• A methodological approach for semi-automatically identifying POEC was established.
• A rule-based algorithm for POEC detection was developed and trained on annotated EC data.
• A POEC library was created, suitable for evaluating EHR-based CTRSS.
• The PENELOPE-C2Q pipeline and PENELOPE POEC library were introduced to support EHR data reuse.

Conclusions:

• The proposed methodology enables semi-automatic detection and standardization of POEC.
• The POEC library provides a valuable resource for developing and evaluating CTRSS.
• This approach can enhance the efficiency of patient-trial matching and facilitate EHR data reuse in clinical research.