Guidelines For Rigorous Evaluation of Clinical LLMs For Conversational Reasoning
Johri, S.; Jeong, J.; Tran, B. A.; Schlessinger, D. I.; Wongvibulsin, S.; Cai, Z. R.; Daneshjou, R.; Rajpurkar, P.
Show abstract
The integration of Large Language Models (LLMs) like GPT-4 and GPT-3.5 into clinical diagnostics has the potential to transform patient-doctor interactions. However, the readiness of these models for real-world clinical application remains inadequately tested. This paper introduces the Conversational Reasoning Assessment Framework for Testing in Medicine (CRAFT-MD), a novel approach for evaluating clinical LLMs. Unlike traditional methods that rely on structured medical exams, CRAFT-MD focuses on natural dialogues, using simulated AI agents to interact with LLMs in a controlled, ethical environment. We applied CRAFT-MD to assess the diagnostic capabilities of GPT-4 and GPT-3.5 in the context of skin diseases. Our experiments revealed critical insights into the limitations of current LLMs in terms of clinical conversational reasoning, history taking, and diagnostic accuracy. Based on these findings, we propose a comprehensive set of guidelines for future evaluations of clinical LLMs. These guidelines emphasize realistic doctor-patient conversations, comprehensive history taking, open-ended questioning, and a combination of automated and expert evaluations. The introduction of CRAFT-MD marks a significant advancement in LLM testing, aiming to ensure that these models augment medical practice effectively and ethically.
Matching journals
The top 8 journals account for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
Similar papers in this journal
- Evaluation of a clinical decision support system for detection of patients at risk after kidney transplantation 91%
- Applying machine-learning to rapidly analyse large qualitative text datasets to inform the COVID-19 pandemic response: Comparing human and machine-assisted topic analysis techniques 90%
- AI-Driven Early Detection of Severe Influenza in Jiangsu, China: A Deep Learning Model Validated Through The Design of Multi-Center Clinical Trials and Prospective Real-World Deployment 90%
Similar papers in this journal
- Protocol For Human Evaluation of Artificial Intelligence Chatbots in Clinical Consultations 93%
- ChatGPT- versus human-generated answers to frequently asked questions about diabetes: a Turing test-inspired survey among employees of a Danish diabetes center 92%
- Robust Disease Prognosis via Diagnostic Knowledge Preservation: A Sequential Learning Approach 91%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.