Papers › Towards Expert-Level Medical Question Answering with Large Language Models

Towards Expert-Level Medical Question Answering with Large Language Models

16 May 2023arXiv:2305.09617archive 2025-07-28

Karan Singhal, Tao Tu, Juraj Gottweis, Rory Sayres, Ellery Wulczyn, Le Hou, Kevin Clark, Stephen Pfohl, Heather Cole-Lewis, Darlene Neal, Mike Schaekermann, Amy Wang, Mohamed Amin, Sami Lachgar, Philip Mansfield, Sushant Prakash, Bradley Green, Ewa Dominowska, Blaise Aguera y Arcas, Nenad Tomasev, Yun Liu, Renee Wong, Christopher Semturs, S. Sara Mahdavi, Joelle Barral, Dale Webster, Greg S. Corrado, Yossi Matias, Shekoofeh Azizi, Alan Karthikesalingam, Vivek Natarajan

Recent artificial intelligence (AI) systems have reached milestones in "grand challenges" ranging from Go to protein-folding. The capability to retrieve medical knowledge, reason over it, and answer medical questions comparably to physicians has long been viewed as one such grand challenge. Large language models (LLMs) have catalyzed significant progress in medical question answering; Med-PaLM was the first model to exceed a "passing" score in US Medical Licensing Examination (USMLE) style questions with a score of 67.2% on the MedQA dataset. However, this and other prior work suggested significant room for improvement, especially when models' answers were compared to clinicians' answers. Here we present Med-PaLM 2, which bridges these gaps by leveraging a combination of base LLM improvements (PaLM 2), medical domain finetuning, and prompting strategies including a novel ensemble refinement approach. Med-PaLM 2 scored up to 86.5% on the MedQA dataset, improving upon Med-PaLM by over 19% and setting a new state-of-the-art. We also observed performance approaching or exceeding state-of-the-art across MedMCQA, PubMedQA, and MMLU clinical topics datasets. We performed detailed human evaluations on long-form questions along multiple axes relevant to clinical applications. In pairwise comparative ranking of 1066 consumer medical questions, physicians preferred Med-PaLM 2 answers to those produced by physicians on eight of nine axes pertaining to clinical utility (p < 0.001). We also observed significant improvements compared to Med-PaLM on every evaluation axis (p < 0.001) on newly introduced datasets of 240 long-form "adversarial" questions to probe LLM limitations. While further studies are necessary to validate the efficacy of these models in real-world settings, these results highlight rapid progress towards physician-level performance in medical question answering.

PaperPDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

m42-health/med42 mentioned on GitHubMIT report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

MMLUMultiple Choice Question Answering (MCQA)Protein FoldingQuestion Answering

2 archive task tags without a task page not shown.

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Multiple Choice Question Answering (MCQA) MMLU (Anatomy) Med-PaLM 2 (ER) Accuracy 84.4 #1 of 3 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MMLU (Anatomy) Med-PaLM 2 (CoT + SC) Accuracy 80.0 #2 of 3 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MMLU (Anatomy) Med-PaLM 2 (5-shot) Accuracy 77.8 #3 of 3 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MMLU (Clinical Knowledge) Med-PaLM 2 (ER) Accuracy 88.7 #1 of 3 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MMLU (Clinical Knowledge) Med-PaLM 2 (5-shot) Accuracy 88.3 #2 of 3 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MMLU (Clinical Knowledge) Med-PaLM 2 (CoT + SC) Accuracy 88.3 #3 of 3 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MMLU (College Biology) Med-PaLM 2 (ER) Accuracy 95.8 #1 of 8 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MMLU (College Biology) Med-PaLM 2 (CoT + SC) Accuracy 95.1 #2 of 8 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MMLU (College Biology) Med-PaLM 2 (5-shot) Accuracy 94.4 #3 of 8 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MMLU (College Medicine) Med-PaLM (ER) Accuracy 83.2 #1 of 3 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MMLU (College Medicine) Med-PaLM (CoT + SC) Accuracy 81.5 #2 of 3 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MMLU (College Medicine) Med-PaLM 2 (5-shot) Accuracy 80.9 #3 of 3 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MMLU (Medical Genetics) Med-PaLM 2 (ER) Accuracy 92 #1 of 8 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MMLU (Medical Genetics) Med-PaLM 2 (5-shot) Accuracy 90 #2 of 8 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MMLU (Medical Genetics) Med-PaLM 2 (CoT + SC) Accuracy 89 #3 of 8 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MMLU (Professional medicine) Med-PaLM 2 (5-shot) Accuracy 95.2 #1 of 6 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MMLU (Professional medicine) Med-PaLM 2 (CoT + SC) Accuracy 93.4 #2 of 6 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MMLU (Professional medicine) Med-PaLM 2 (ER) Accuracy 92.3 #3 of 6 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MedMCQA Med-PaLM 2 (ER) Test Set (Acc-%) 0.723 #1 of 22 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MedMCQA Med-PaLM 2 (CoT+SC) Test Set (Acc-%) 0.715 #2 of 22 Archive leaderboard report
Multiple Choice Question Answering (MCQA) MedMCQA Med-PaLM 2 (5-shot) Test Set (Acc-%) 0.713 #3 of 22 Archive leaderboard report
Question Answering MedQA Med-PaLM 2 Accuracy 85.4 #3 of 27 Archive leaderboard report
Question Answering MedQA Med-PaLM 2 (CoT + SC) Accuracy 83.7 #4 of 27 Archive leaderboard report
Question Answering MedQA Med-PaLM 2 (5-shot) Accuracy 79.7 #5 of 27 Archive leaderboard report
Question Answering PubMedQA Med-PaLM 2 (5-shot) Accuracy 79.2 #4 of 30 Archive leaderboard report
Question Answering PubMedQA Med-PaLM 2 (ER) Accuracy 75.0 #17 of 30 Archive leaderboard report
Question Answering PubMedQA Med-PaLM 2 (CoT + SC) Accuracy 74.0 #19 of 30 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

BASE

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections