{"id":"W3183267169","doi":"10.1093/gigascience/giab055","title":"Preventing dataset shift from breaking machine-learning biomarkers","year":2021,"lang":"en","type":"review","venue":"GigaScience","topic":"Artificial Intelligence in Healthcare and Education","field":"Medicine","cited_by":92,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"National Institute of Biomedical Imaging and Bioengineering; National Institute of Mental Health; National Institutes of Health; Agence Nationale de la Recherche","keywords":"Biomarker; Machine learning; Artificial intelligence; Computer science; Biomarker discovery; Population; Medicine; Biology; Proteomics","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":[],"consensus_categories":[],"category_scores_codex":[0.00087161,0.0002703747,0.0009072427,0.0001788425,0.000325031,0.0001239721,0.0003425509,0.0002092355,0.0004676149],"category_scores_gemma":[0.001000737,0.0002297151,0.0002411701,0.0008431497,0.0001259784,0.0001666194,0.000171968,0.0006533268,0.0002432127],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001664546,"about_ca_system_score_gemma":0.001182619,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002979937,"about_ca_topic_score_gemma":0.0001183133,"domain_scores_codex":[0.9974453,0.0002356268,0.0007292185,0.0007475629,0.0004096046,0.0004326222],"domain_scores_gemma":[0.9983498,0.0004020189,0.0003931587,0.0005670427,0.0000677442,0.00022026],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.000001459365,0.00002981833,0.0003650288,0.00200986,0.00003911011,0.00003898804,0.0002460772,5.272836e-7,0.000002590205,0.00001187861,0.0007443375,0.9965103],"study_design_scores_gemma":[0.00001022357,0.00003148293,0.00002673112,0.01700548,0.0003664939,0.0000669338,0.0001998697,0.0001298156,0.00002147999,0.00003791563,0.981895,0.000208578],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"review","genre_gemma":"review","genre_scores_codex":[0.0001572102,0.9974869,0.0002750795,0.0001802307,0.0008611511,0.0004330798,0.0003007373,0.00005359635,0.0002520106],"genre_scores_gemma":[0.0003573782,0.9903492,0.001344294,0.0002198505,0.0004679065,0.00004149135,0.007056036,0.00003168293,0.0001321377],"genre_candidate":"review","genre_consensus":"review","teacher_disagreement_score":0.9963018,"threshold_uncertainty_score":0.9367508,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.2725123400981636,"score_gpt":0.4901105873463265,"score_spread":0.2175982472481629,"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."}}