{"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":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.09332503,0.001175408,0.002344309,0.00437691,0.002497799,0.007689379,0.005301422,0.007192334,0.014457],"category_scores_gemma":[0.3297296,0.001512727,0.002149542,0.003666921,0.01222575,0.0168761,0.005953307,0.007838337,0.0008561997],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.003854355,"about_ca_system_score_gemma":0.004120873,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004149787,"about_ca_topic_score_gemma":0.002850583,"domain_scores_codex":[0.9543545,0.03005007,0.002581702,0.006317382,0.005206487,0.001489843],"domain_scores_gemma":[0.4916467,0.4437731,0.02303328,0.02806305,0.01073316,0.002750674],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0001757863,0.00007483759,0.008487291,0.0003582718,0.0002729167,0.0003111413,0.0005726056,0.005649052,0.0002925587,0.9597141,0.002798728,0.02129277],"study_design_scores_gemma":[0.00006521827,0.00004396028,0.0009817794,0.0001214996,0.0001009429,0.0002175597,0.0001087559,0.01602072,0.000456196,0.9785122,0.003343047,0.00002804768],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.03229078,0.007487736,0.8607756,0.07986053,0.001589944,0.0003438262,0.001083605,0.0004724095,0.01609554],"genre_scores_gemma":[0.8524902,0.004407584,0.1159804,0.0114867,0.005994658,0.0005165071,0.0006579229,0.0004159119,0.008049989],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.906675,"threshold_uncertainty_score":0.4935558,"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."}}