cs.LGNov 6, 2026

TrialJev: Open-Weight Single Pass Eligibility Scoring for Patient Trial Screening

Authors: Omar Ibrahim, Grant Hamblin, Benjamin Haibe-Kains

Organizations: AI Hub, University Health Network (UHN), Toronto, ON · AI Hub, University Health Network (UHN), Toronto, ON, Canada · Department of Medical Biophysics, University of Toronto, Toronto, ON, Canada · Vector Institute for Artificial Intelligence, Toronto, ON, Canada · 

Abstract

Clinical trial screening requires assessing large numbers of patient-trial pairs to prioritize candidates for detailed eligibility review, often under restrictions on sharing patient records. We introduce TrialJev, an open-weight eligibility scorer that uses LoRA-adapted Qwen models and a pointer head to score natural-language answer options in a single forward pass, without autoregressive decoding. Trained on TREC Clinical Trials 2021 and evaluated on 6,975 eligible-or-excluded pairs from 50 unseen TREC 2022 synthetic patients, TrialJev achieves 80.97% (Qwen3.5-27B) and 81.84% (Qwen3.8-27B) accuracy, compared with 79.87% and 79.93% for LLMs fine-tuned on the same data to generate verdicts. In matched batch-1 measurements, verdict generation takes 2.45-2.70 times as long as the one-pass inference of TrialJev at 9B-27B. Under a common development-fitted decision correction, the two 27B TrialJev models reach 80.99% and 81.88% accuracy, compared with 81.28% for the proprietary hosted Jev API; paired confidence intervals include zero. TrialJev runs locally on a single GPU, requiring approximately 178 ms per pair at 27B, without transmitting patient text to an external model API. An A/B-logit ablation shows that avoiding decoding, rather than the pointer head alone, accounts for the measured efficiency advantage. TrialJev is intended for first-pass prioritization of candidate trials, complementing rather than replacing explanatory models and clinical review. Evaluation is limited to synthetic patients and implementation-specific timing, and clinical use has not been validated. Adapters, pointer heads, and model cards are released on Hugging Face.

Figures & tables

Appendix figures & tables25 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. Can JEV Support Clinical Trial Criteria Assessment?

    Oct 26, 2026Mengxian He, Ziyue Gao, Chun Li +4Clinical Decision SupportClinical Language Model Evaluation

  2. Decision language models as affordable and rapid alternatives to LLM judges of clinical generative AI

    Oct 20, 2026Rotem Gershon, Seffi Cohen, Dean Ariel +11LLM-as-a-JudgeClinical Language Model Evaluation

  3. RECITE: Revising Eligibility Criteria Incorporating Textual Evidence

    Oct 30, 2026Russell Ro, Reza Abbasi-AslClinical Information ExtractionEvidence Retrieval