{"id":"W4391843481","doi":"10.1038/s42256-024-00797-8","title":"A causal perspective on dataset bias in machine learning for medical imaging","year":2024,"lang":"en","type":"article","venue":"Nature Machine Intelligence","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":68,"is_retracted":false,"has_abstract":false,"ca_institutions":"Hospital for Sick Children","funders":"Engineering and Physical Sciences Research Council; Royal Academy of Engineering; Alan Turing Institute; Microsoft Research","keywords":"Perspective (graphical); Computer science; Artificial intelligence; Machine learning; Data science; Presentation (obstetrics); Causal inference; Causal structure; Debiasing; Causal model; Risk analysis (engineering); Psychology; Medicine; Cognitive science","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":"codex-gemma-dda1882f352a","candidate_categories":["research_integrity"],"consensus_categories":[],"category_scores_codex":[0.00125108,0.000256807,0.000311608,0.0004983161,0.0001239629,0.00007185216,0.0002219462,0.0003296605,0.0006179652],"category_scores_gemma":[0.005635348,0.0002071576,0.0001131359,0.0006685321,0.0001067922,0.0001303916,0.00005398054,0.002899996,0.0001791646],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004749734,"about_ca_system_score_gemma":0.0005001191,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003581484,"about_ca_topic_score_gemma":0.001544767,"domain_scores_codex":[0.9976676,0.0001039853,0.0005434866,0.0006720839,0.0005710929,0.0004418162],"domain_scores_gemma":[0.9977488,0.001455203,0.0000580918,0.0002931585,0.0001983982,0.0002463388],"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":[0.001347204,0.0006449654,0.02966413,0.0008232373,0.0001117018,0.00104277,0.006196672,0.0005541384,0.0002452075,0.06040037,0.0194843,0.8794853],"study_design_scores_gemma":[0.0001579029,0.001004106,0.001111036,0.00252626,0.0001154611,0.0006550731,0.004686118,0.7668163,0.01702197,0.02103138,0.1841965,0.0006779322],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"commentary","genre_gemma":"empirical","genre_scores_codex":[0.1410017,0.1867135,0.1853152,0.4397344,0.01979257,0.008599316,0.002615544,0.002006274,0.01422146],"genre_scores_gemma":[0.9930117,0.0005907185,0.0004013659,0.003498146,0.0008975014,0.00007752131,0.001165925,0.00004768854,0.0003095064],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.8788074,"threshold_uncertainty_score":0.9994004,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.08040236266088432,"score_gpt":0.476153720258939,"score_spread":0.3957513575980547,"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."}}