Medevi newsPublications, conference presentations and project updates from the Medevi team.https://medevi.ca/Medevi: Large Language Model-Assisted Critical Appraisal for Medical Education Researchhttps://medevi.ca/news/2026-icam-medevi-critical-appraisal/https://medevi.ca/news/2026-icam-medevi-critical-appraisal/LLM-assisted appraisal reached substantial agreement with human reviewers on MERSQI (ICC 0.755) at under a minute and $0.10–$0.57 per paper, while NOS-E agreement remained fair (ICC 0.345).Wed, 01 Apr 2026 00:00:00 GMTPresentationMedevi-RoB: Benchmarking Large Language Models for Automated Risk-of-Bias (RoB 2) Assessmenthttps://medevi.ca/news/2025-tcairem-medevi-rob/https://medevi.ca/news/2025-tcairem-medevi-rob/Open reasoning models with structured prompting reached human-level agreement on all five RoB 2 domains across 273 outcome-level judgements, substantially outperforming earlier GPT-4 baselines.Sat, 01 Nov 2025 00:00:00 GMTPresentationProtocol for Cross-Domain Benchmark for Automated Risk of Bias 2 (RoB 2) Assessment in Randomized Trials using Large Language Models (Medevi-RoB)https://medevi.ca/news/2025-medevi-rob-protocol/https://medevi.ca/news/2025-medevi-rob-protocol/Pre-registered protocol for Medevi-RoB, a cross-domain benchmark of large language models for automated Risk of Bias 2 assessment of randomized trials against Cochrane expert consensus.Mon, 15 Sep 2025 00:00:00 GMTPublication · Protocol