Researchers have introduced MedJev, an open-source System One model designed for efficient clinical decision support. This model operates in a single forward pass without generating text, making it computationally efficient. MedJev was evaluated on ten medical benchmarks, including knowledge, diagnosis, and patient-level prediction tasks, using datasets like MIMIC-IV. It demonstrated competitive performance against zero-shot LLM baselines and supervised models, achieving high accuracies on benchmarks such as MedQA and USMLE, while also showing good confidence calibration. The model also showed promise in patient-level predictions, outperforming XGBoost on in-hospital mortality prediction, though it required recalibration for local patient populations. AI
IMPACT Provides a promising open-source baseline for efficient, low-cost clinical decision support systems.
RANK_REASON The cluster describes a research paper introducing a new model and its evaluation on medical benchmarks. [lever_c_demoted from research: ic=1 ai=1.0]
- GPT-4
- Jevíčko
- MedQA
- MIMIC-IV
- System One
- The New England Journal of Medicine
- United States Medical Licensing Examination
- XGBoost
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
