{"id":"W4405499503","doi":"10.1148/radiol.241711","title":"Methodological Challenges in Evaluating Large Language Models in Radiology","year":2024,"lang":"en","type":"letter","venue":"Radiology","topic":"Radiomics and Machine Learning in Medical Imaging","field":"Medicine","cited_by":1,"is_retracted":false,"has_abstract":false,"ca_institutions":"Western University","funders":"","keywords":"Medicine; Medical physics; Radiology","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":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.1099254,0.0005809463,0.002097023,0.001437495,0.001899457,0.008992412,0.004562132,0.007339748,0.004627168],"category_scores_gemma":[0.5097129,0.000871776,0.001135946,0.001570153,0.004496868,0.006061281,0.003088245,0.01073219,0.003843874],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003157623,"about_ca_system_score_gemma":0.00507756,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01002681,"about_ca_topic_score_gemma":0.01709402,"domain_scores_codex":[0.9071251,0.07644421,0.005179628,0.002379583,0.008153033,0.0007185772],"domain_scores_gemma":[0.4030907,0.5483549,0.005576101,0.01396262,0.02655156,0.002464052],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"not_applicable","study_design_scores_codex":[0.001429928,0.0003982346,0.0223425,0.0009756166,0.0007385187,0.002013509,0.001444965,0.01927815,0.002019705,0.06849483,0.5271137,0.3537503],"study_design_scores_gemma":[0.0006739375,0.0002964145,0.008075266,0.0006987121,0.0003350904,0.004514399,0.001902991,0.1829119,0.003268125,0.6461351,0.1509585,0.0002296187],"study_design_candidate":"not_applicable","study_design_consensus":"not_applicable","genre_codex":"commentary","genre_gemma":"commentary","genre_scores_codex":[0.01449147,0.004556225,0.1454619,0.8154102,0.00920065,0.0002282703,0.002011755,0.001015033,0.007624629],"genre_scores_gemma":[0.3779505,0.004850144,0.3201554,0.2397406,0.0448269,0.001295473,0.002101863,0.001454288,0.007624743],"genre_candidate":"commentary","genre_consensus":"commentary","teacher_disagreement_score":0.8900746,"threshold_uncertainty_score":0.5813481,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.2032164414098568,"score_gpt":0.4412040421299666,"score_spread":0.2379876007201098,"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."}}