{"id":"W4414499338","doi":"10.1056/aidbp2500120","title":"Assessment of Large Language Models in Clinical Reasoning: A Novel Benchmarking Study","year":2025,"lang":"en","type":"article","venue":"NEJM AI","topic":"Clinical Reasoning and Diagnostic Skills","field":"Medicine","cited_by":28,"is_retracted":false,"has_abstract":false,"ca_institutions":"University of Alberta","funders":"","keywords":"Benchmarking; Language model; Quality (philosophy); Identification (biology); Measure (data warehouse)","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.03285042,0.001176179,0.000924357,0.002170961,0.00100646,0.003831927,0.002854826,0.002115275,0.004176596],"category_scores_gemma":[0.1691909,0.0004690127,0.001314042,0.002020314,0.001310194,0.006518242,0.004152421,0.002280743,0.0009169761],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002375114,"about_ca_system_score_gemma":0.002506743,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005561337,"about_ca_topic_score_gemma":0.005595571,"domain_scores_codex":[0.9691153,0.02262321,0.002062077,0.002534165,0.003160886,0.0005044734],"domain_scores_gemma":[0.7965477,0.1663193,0.004686168,0.01727273,0.01299838,0.002175751],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.01109319,0.01051299,0.1610761,0.002197625,0.002070023,0.0008973961,0.006523888,0.2370465,0.009432373,0.02393337,0.01963533,0.5155813],"study_design_scores_gemma":[0.001331806,0.004583814,0.03695701,0.0004134752,0.0007142548,0.0007398098,0.002705432,0.8999142,0.007938802,0.03420152,0.01032613,0.0001737381],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8368357,0.001664666,0.1446742,0.001337853,0.0001739121,0.0008842028,0.002644619,0.001717593,0.01006729],"genre_scores_gemma":[0.939348,0.0002142243,0.05490075,0.0002796171,0.00004803904,0.0003906329,0.003673323,0.0002578735,0.000887538],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.03285042,"threshold_uncertainty_score":0.1737317,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03709534502479742,"score_gpt":0.4589540775897755,"score_spread":0.4218587325649781,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}