Truth, Trust, and Trouble: Medical AI on the Edge
Not on the map. Why?
Abstract (arXiv)
Large Language Models (LLMs) hold significant promise for transforming digital health by enabling automated medical question answering. However, ensuring these models meet critical industry standards for factual accuracy, usefulness, and safety remains a challenge, especially for open-source solutions. We present a rigorous benchmarking framework using a dataset of over 1,000 health questions. We assess model performance across honesty, helpfulness, and harmlessness. Our results highlight trade-offs between factual reliability and safety among evaluated models -- Mistral-7B, BioMistral-7B-DARE, and AlpaCare-13B. AlpaCare-13B achieves the highest accuracy (91.7%) and harmlessness (0.92), while domain-specific tuning in BioMistral-7B-DARE boosts safety (0.90) despite its smaller scale. Few-shot prompting improves accuracy from 78% to 85%, and all models show reduced helpfulness on complex queries, highlighting ongoing challenges in clinical QA.
Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 8 of its 24 references that are in the map. First posted on arXiv 1 Jul 2025.
Builds on
8 of 24 references.
References · 8
- 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,626
- 2021TruthfulQA: Measuring How Models Mimic Human Falsehoodscited by 2,975
- 2022Large Language Models Encode Clinical Knowledgecited by 1,830
- 2019PubMedQA: A Dataset for Biomedical Research Question Answeringcited by 1,288
- 2024BioMistral: A Collection of Open-Source Pretrained Large Language Models for Medical Domainscited by 301
- 2023AlpaCare:Instruction-tuned Large Language Models for Medical Applicationcited by 90
- 2025Diagnosing and Addressing Pitfalls in KG-RAG Datasets: Toward More Reliable Benchmarkingcited by 12
- 2024Digital Diagnostics: The Potential Of Large Language Models In Recognizing Symptoms Of Common Illnessescited by 8
Built on by
Citing papers per year: 2026: 2.
2 papers.