{"id":"W4410201002","doi":"10.1007/s10278-025-01523-5","title":"Cross-Institutional Evaluation of Large Language Models for Radiology Diagnosis Extraction: A Prompt-Engineering Perspective","year":2025,"lang":"en","type":"article","venue":"Journal of Imaging Informatics in Medicine","topic":"Radiology practices and education","field":"Medicine","cited_by":4,"is_retracted":false,"has_abstract":false,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"","keywords":"Perspective (graphical); Extraction (chemistry); Computer science; Radiology; Medicine; Artificial intelligence; Chemistry; Chromatography","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":[0.05045153,0.002265851,0.001079036,0.003300643,0.001453732,0.004379053,0.003832884,0.002689464,0.0032238],"category_scores_gemma":[0.100069,0.0008404992,0.002416201,0.002536375,0.002049487,0.006976228,0.005983069,0.003787195,0.002368543],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002187174,"about_ca_system_score_gemma":0.003920508,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01119024,"about_ca_topic_score_gemma":0.01197141,"domain_scores_codex":[0.9673112,0.02230689,0.002806584,0.004169294,0.002538685,0.0008672359],"domain_scores_gemma":[0.8604736,0.1031835,0.003737299,0.01772598,0.01266342,0.002216166],"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":[0.01820038,0.01555932,0.17749,0.004653142,0.008364261,0.002305428,0.00788702,0.1739124,0.03029236,0.008190775,0.06137839,0.4917666],"study_design_scores_gemma":[0.003492938,0.009613475,0.08615081,0.001019418,0.005148621,0.003584872,0.01119733,0.700093,0.09920744,0.02354495,0.05611975,0.0008274003],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8295724,0.003827524,0.1266364,0.004295785,0.0007378759,0.001866987,0.0136386,0.01381088,0.00561359],"genre_scores_gemma":[0.8578168,0.0007525511,0.1023639,0.00114842,0.0001996021,0.0006690141,0.03398436,0.001154381,0.001911073],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.05045153,"threshold_uncertainty_score":0.2668164,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03598449043582892,"score_gpt":0.4389903356725702,"score_spread":0.4030058452367413,"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."}}