Truth, Trust, and Trouble: Medical AI on the Edge

Fuente: arXiv
Enregistré dans:
Détails bibliographiques
Auteurs principaux: Azeez, Mohammad Anas, Ali, Rafiq, Shabbir, Ebad, Siddiqui, Zohaib Hasan, Kashyap, Gautam Siddharth, Gao, Jiechao, Naseem, Usman
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866918532896260096
author Azeez, Mohammad Anas
Ali, Rafiq
Shabbir, Ebad
Siddiqui, Zohaib Hasan
Kashyap, Gautam Siddharth
Gao, Jiechao
Naseem, Usman
author_facet Azeez, Mohammad Anas
Ali, Rafiq
Shabbir, Ebad
Siddiqui, Zohaib Hasan
Kashyap, Gautam Siddharth
Gao, Jiechao
Naseem, Usman
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2507_02983
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Truth, Trust, and Trouble: Medical AI on the Edge
Azeez, Mohammad Anas
Ali, Rafiq
Shabbir, Ebad
Siddiqui, Zohaib Hasan
Kashyap, Gautam Siddharth
Gao, Jiechao
Naseem, Usman
Computation and Language
Artificial Intelligence
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.
title Truth, Trust, and Trouble: Medical AI on the Edge
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2507.02983