{"doi":"10.1038/s41591-024-03423-7","title":"Toward expert-level medical question answering with large language models","abstract":"Large language models (LLMs) have shown promise in medical question answering, with Med-PaLM being the first to exceed a 'passing' score in United States Medical Licensing Examination style questions. However, challenges remain in long-form medical question answering and handling real-world workflows. Here, we present Med-PaLM 2, which bridges these gaps with a combination of base LLM improvements, medical domain fine-tuning and new strategies for improving reasoning and grounding through ensemble refinement and chain of retrieval. Med-PaLM 2 scores up to 86.5% on the MedQA dataset, improving upon Med-PaLM by over 19%, and demonstrates dramatic performance increases across MedMCQA, PubMedQA and MMLU clinical topics datasets. Our detailed human evaluations framework shows that physicians prefer Med-PaLM 2 answers to those from other physicians on eight of nine clinical axes. Med-PaLM 2 also demonstrates significant improvements over its predecessor across all evaluation metrics, particularly on new adversarial datasets designed to probe LLM limitations (P < 0.001). In a pilot study using real-world medical questions, specialists preferred Med-PaLM 2 answers to generalist physician answers 65% of the time. While specialist answers were still preferred overall, both specialists and generalists rated Med-PaLM 2 to be as safe as physician answers, demonstrating its growing potential in real-world medical applications.","journal":"Nature Medicine","year":2025,"id":508703,"datarank":10.33461046809183,"base_score":6.677083461247136,"endowment":6.677083461247136,"self_citation_contribution":1.0015625191870705,"citation_network_contribution":9.33304794890476,"self_endowment_contribution":1.0015625191870705,"citer_contribution":9.33304794890476,"corpus_percentile":null,"corpus_rank":null,"citation_count":793,"citer_count":200,"citers_with_citation_signal":200,"citers_with_endowment":200,"datacite_reuse_total":0,"is_dataset":false,"is_dataset_confidence":0.9586,"is_data_producer":false,"deposit_databanks":null,"is_oa":true,"file_count":0,"downloads":0,"has_version_chain":false,"published_date":"2025-01-01","fair_score":null,"fair_percentile":null,"algorithm_id":"datarank_citation_only_1hop_v6","ranking_scope":"data_only","authors":[{"id":686407,"name":"Tao Tu","orcid":"0000-0001-9191-7938","position":1,"is_corresponding":false},{"id":1362042,"name":"Juraj Gottweis","orcid":null,"position":2,"is_corresponding":false},{"id":1201667,"name":"Rory Sayres","orcid":"0000-0002-8269-5779","position":3,"is_corresponding":false},{"id":1362043,"name":"Ellery Wulczyn","orcid":null,"position":4,"is_corresponding":false},{"id":1360744,"name":"Mohamed Amin","orcid":"0009-0002-4874-8272","position":5,"is_corresponding":false},{"id":273237,"name":"Le Hou","orcid":null,"position":6,"is_corresponding":false},{"id":1118255,"name":"Kevin Clark","orcid":"0000-0003-4974-6885","position":7,"is_corresponding":false},{"id":268682,"name":"Stephen Pfohl","orcid":"0000-0003-0551-9664","position":8,"is_corresponding":false},{"id":1171999,"name":"Heather Cole-Lewis","orcid":"0000-0002-7275-1810","position":9,"is_corresponding":false},{"id":1202260,"name":"Darlene Neal","orcid":null,"position":10,"is_corresponding":false},{"id":1202262,"name":"Qazi Mamunur Rashid","orcid":null,"position":11,"is_corresponding":false},{"id":419880,"name":"Mike Schaekermann","orcid":"0000-0002-1735-9680","position":12,"is_corresponding":false},{"id":467527,"name":"Amy Wang","orcid":"0009-0004-6846-786X","position":13,"is_corresponding":false},{"id":973572,"name":"Dev Dash","orcid":null,"position":14,"is_corresponding":false},{"id":227523,"name":"Jonathan H. Chen","orcid":"0000-0002-4387-8740","position":15,"is_corresponding":false},{"id":226836,"name":"Nigam H. Shah","orcid":"0000-0001-9385-7158","position":16,"is_corresponding":false},{"id":1360745,"name":"Sami Lachgar","orcid":"0009-0008-3663-1441","position":17,"is_corresponding":false},{"id":1201674,"name":"P. Mansfield","orcid":"0000-0003-4969-0543","position":18,"is_corresponding":false},{"id":1201675,"name":"Sushant Prakash","orcid":"0009-0000-4162-4600","position":19,"is_corresponding":false},{"id":1360746,"name":"Bradley Green","orcid":"0000-0001-9589-0226","position":20,"is_corresponding":false},{"id":1360747,"name":"Ewa Dominowska","orcid":"0009-0006-5644-9685","position":21,"is_corresponding":false},{"id":624049,"name":"Blaise Agüera y Arcas","orcid":"0000-0003-2256-9823","position":22,"is_corresponding":false},{"id":1201668,"name":"Nenad Tomašev","orcid":"0000-0003-1624-0220","position":23,"is_corresponding":false},{"id":661611,"name":"Yun Liu","orcid":"0000-0003-4079-8275","position":24,"is_corresponding":false},{"id":1360748,"name":"Renee Wong","orcid":"0009-0003-0403-7679","position":25,"is_corresponding":false},{"id":1035508,"name":"Christopher Semturs","orcid":"0000-0001-6108-2773","position":26,"is_corresponding":false},{"id":1161715,"name":"S. Sara Mahdavi","orcid":"0000-0001-6823-598X","position":27,"is_corresponding":false},{"id":1201677,"name":"Joëlle Barral","orcid":"0009-0009-0432-5148","position":28,"is_corresponding":false},{"id":620996,"name":"Dale R. Webster","orcid":"0000-0002-3023-8824","position":29,"is_corresponding":false},{"id":58456,"name":"Greg S. Corrado","orcid":"0000-0001-8817-0992","position":30,"is_corresponding":false},{"id":551685,"name":"Yossi Matias","orcid":"0000-0003-3960-6002","position":31,"is_corresponding":false},{"id":1201669,"name":"Shekoofeh Azizi","orcid":"0000-0002-7447-6031","position":32,"is_corresponding":false},{"id":1201676,"name":"Alan Karthikesalingam","orcid":"0009-0000-4958-5976","position":33,"is_corresponding":false},{"id":1201685,"name":"Vivek Natarajan","orcid":"0000-0001-7849-2074","position":34,"is_corresponding":false},{"id":1201679,"name":"K. K. Singhal","orcid":"0009-0001-0286-609X","position":0,"is_corresponding":true}],"reference_count":57,"raw_metadata":{"citation_network_status":"fetched"},"created_at":"2026-07-19T02:47:00.188082Z","pmid":"39779926","pmcid":null,"fwci":null,"citation_percentile":null,"influential_citations":0,"oa_status":null,"license":null,"views":0,"total_file_size_bytes":0,"version_count":0,"fair_f":null,"fair_a":null,"fair_i":null,"fair_r":null,"fair_zscore":null,"fair_rationale":null,"fair_model":null,"fair_agent_version":null,"fair_fulltext_source":null,"fair_has_llm":null,"fair_computed_at":null,"clinical_trials":[],"software_tools":[],"db_accessions":[],"linked_datasets":[],"topics":[]}