importance 5/10ai assistedArtificial Intelligence in Healthcare and Educationclinical trialsstage: researchmixed
Problem
Matching patients to clinical trials requires analyzing eligibility against complex inclusion and exclusion criteria, which is slow and hard to scale.
Why it matters
Poor trial matching slows enrollment and limits patient access to new treatments.
What the AI did
LLMs analyzed patient eligibility against trial criteria; open-source LLMs were fine-tuned on a GPT-4-generated synthetic dataset and evaluated with automated and human-centric assessments.
Results
Fine-tuned open-source LLMs achieved performance parity with proprietary counterparts such as GPT-3.5.
Impact
Evidence supports LLM feasibility for trial matching, with fine-tuned open-source models matching proprietary ones.
Record
Venue
Journal of the American Medical Informatics Association
large language models, clinical trial matching, fine-tuning, GPT-4, LLaMA, eligibility
Importance rating
5/ 10
real but narrower advance
Trial matching is called high-stakes in healthcare, but abstract quantifies no patient or trial scale; feasibility demonstration with fine-tuned open-source models, no deployment.
Basis: inferred.
Judged from the source text, which gives no explicit scale.
Stated scale: not stated in source. Method and limits.