{"id":"W3126829124","doi":"10.1038/s42003-021-01674-5","title":"Systematic auditing is essential to debiasing machine learning in biology","year":2021,"lang":"en","type":"article","venue":"Communications Biology","topic":"Cell Image Analysis Techniques","field":"Biochemistry, Genetics and Molecular Biology","cited_by":40,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"National Institute of Mental Health; Stanley Center for Psychiatric Research, Broad Institute; University of California, San Francisco; Simons Foundation Autism Research Initiative; Broad Institute; McGill University; Harvard University; Simons Foundation; H. Lundbeck A/S; Lundbeckfonden; U.S. Department of Health and Human Services","keywords":"Debiasing; Audit; Computer science; Process (computing); Data science; Machine learning; Artificial intelligence; Code (set theory); Psychology; Accounting; Cognitive science","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004254754,0.0001297512,0.0002947611,0.0001450104,0.0001479915,0.00002296863,0.0006336055,0.0001730239,0.00003690659],"category_scores_gemma":[0.001103005,0.0001348652,0.0000981469,0.0003450504,0.0001134667,0.000003330591,0.001039709,0.0002081608,0.00002770543],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000313925,"about_ca_system_score_gemma":0.00006712612,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001091297,"about_ca_topic_score_gemma":0.0008457148,"domain_scores_codex":[0.9982101,0.000764008,0.0004213794,0.0003454609,0.00002954576,0.0002294553],"domain_scores_gemma":[0.9980772,0.0001190618,0.0001374085,0.001463917,0.0001556753,0.00004679132],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.000005363595,0.00007237764,0.01412975,0.0001177958,0.00008091167,0.000002077611,0.0001268796,0.00001221296,0.9831633,0.0006430984,0.0002051493,0.001441073],"study_design_scores_gemma":[0.0006442033,0.0002640226,0.0008707052,0.0006982474,0.0001520832,0.0001136082,0.0006370727,0.005351353,0.9354714,0.001386297,0.05366423,0.0007468262],"study_design_candidate":"bench_or_experimental","study_design_consensus":"bench_or_experimental","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8446683,0.03918513,0.09172782,0.007817662,0.0001251255,0.001107875,0.00003658768,0.0002392191,0.01509225],"genre_scores_gemma":[0.9843722,0.00123027,0.01231522,0.0008580133,0.00003853706,0.00008602023,0.0007079538,0.00001759111,0.0003741731],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1397039,"threshold_uncertainty_score":0.5499644,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01569015251213753,"score_gpt":0.3388596447208553,"score_spread":0.3231694922087178,"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."}}