Enhancing Patient-Trial Matching With Large Language Models: A Scoping Review of Emerging Applications and Approaches

Abstract

Purpose: Patient recruitment remains a major bottleneck in clinical trial execution, with inefficient patient-trial matching often causing delays and failures. Recent advancements in large language models (LLMs) offer a promising avenue for automating and improving this process. This scoping review aims to provide a comprehensive synthesis of the emerging applications of LLMs in patient-trial matching.

Methods: A comprehensive search was conducted in PubMed, Web of Science, and OpenAlex for literature published between December 1, 2022, and December 31, 2024. Studies were included if they explicitly integrated LLMs into patient-trial matching systems. Data extraction focused on system architectures, patient data processing, eligibility criteria processing, matching techniques, evaluation metrics, and performance.

Results: Of the 2,357 studies initially identified, 24 met the inclusion criteria. The majority (21/24) were published in 2024, highlighting the rapid adoption of LLMs in this domain. Most systems used patient-centric matching (17/24), with OpenAI's generative pretrained transformer models being the most commonly used LLM. Core components of these systems included eligibility criteria processing, patient data processing, and matching, with some incorporating retrieval algorithms to enhance computational efficiency. LLM-integrated approaches demonstrated improved accuracy and scalability in patient-trial matching, although challenges such as performance variability, interpretability, and reliance on synthetic data sets remain significant.

Conclusion: LLM-based patient-trial matching systems present a transformative opportunity to enhance the efficiency and accuracy of clinical trial recruitment. Despite current limitations related to model generalizability, explainability, and data constraints, future advancements in hybrid modeling strategies, domain-specific fine-tuning, and real-world data set integration could further optimize LLM-based trial matching. Addressing these challenges will be crucial to realizing the full potential of LLMs in streamlining patient recruitment and accelerating clinical trial execution.

FIG 1.

The PRISMA diagram depicts the number of records identified, included and excluded, and the reasons for exclusion. BERT, Bidirectional Encoder Representations from Transformers; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses for Scoping Reviews.

FIG 2.

Popular components of patient-trial matching system. The icons were designed by Freepik.

FIG 3.

The common processing flow for different data types. EHR, electronic health record.

FIG 4.

General architectures of matching components. (A) Matching patients and trials through a single question; (B) matching patients and trials using a series of decomposed questions; (C) generating embeddings as representations for patients or trials to predict matching outcomes; (D) directly matching patients and trials based on extracted keywords. The icons were designed by Freepik. BERT, Bidirectional Encoder Representations from Transformers; KD, Kawasaki disease; LLMs, large language models.

FIG 5.

Common techniques and flows in the matching module. The matching results can be either a class (ie, eligible or not) or a ranking. Matching can occur in a single step or multiple steps. Single-step approaches include executing a query (query), using a neural network to classify based on embeddings (embedding), classifying with an LLM (classification), or scoring with LLMs or retrieval algorithms (scoring). Multistep methods involve a combination of processes such as classification, scoring, and retrieval (retrieval); ensembling multiple classification results for a single sample (ensembling); aggregating unit-level predictions to derive patient-trial–level outcomes (aggregation); and performing exact keyword matching between eligibility criteria and patient profiles (exact matching). LLMs, large language models.