Skye Insights Ltd. is a healthcare AI research firm based in Toronto, Canada. We develop and rigorously evaluate large language models for clinical decision support. Our goal is to give healthcare professionals reliable, evidence-based tools that improve patient outcomes.
Publications
Skyer: a novel benchmark for evaluating the effectiveness of large language models in emergency department triage
Nourbakhsh, S. Canadian Journal of Emergency Medicine (2026).
Two of fifteen large language models benchmarked against human triage experts on 55 realistic pediatric emergency department scenarios matched or exceeded expert accuracy, pointing to a role as clinician-supervised triage assistants.
“From Triage To Discharge”: Do New LLM Versions Handle the ED Better? Skyer Answers.
Nourbakhsh, S., Majzoobi, M. LLM/VLM Deployment Opportunities and Risks in Healthcare (DAIH), workshop at COLM 2026. Peer-reviewed, non-archival.
Skyer generalized into a reusable framework for safety-aware, version-over-version evaluation of large language models on emergency department tasks. Across fifteen GPT, Gemini, and Gemma versions, newer is frequently not better, and an older GPT version outperformed its successor.