{"id":"W3020729718","doi":"10.1016/j.jbi.2020.103424","title":"Building machine learning models without sharing patient data: A simulation-based analysis of distributed learning by ensembling","year":2020,"lang":"en","type":"article","venue":"Journal of Biomedical Informatics","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":39,"is_retracted":false,"has_abstract":false,"ca_institutions":"Alberta Children's Hospital; University of Calgary","funders":"National Institute of Biomedical Imaging and Bioengineering; GE Healthcare; Janssen Research and Development; National Institutes of Health; H. Lundbeck A/S; Johnson and Johnson; Calgary Foundation; Northern California Institute for Research and Education; U.S. Department of Defense; Alzheimer's Disease Neuroimaging Initiative; University of Southern California; Genentech; IXICO; National Institute on Aging; Fujirebio Europe; Heart and Stroke Foundation of Canada; Merck","keywords":"Computer science; Machine learning; Artificial intelligence; Data sharing; Medicine","routes":{"ca_aff":true,"ca_fund":true,"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":[],"consensus_categories":[],"category_scores_codex":[0.005627452,0.0008154584,0.001917456,0.0007116039,0.0009553363,0.001235061,0.00232054,0.002010574,0.002389793],"category_scores_gemma":[0.01765626,0.001064948,0.001329582,0.0005128809,0.001554776,0.001754475,0.001539623,0.002166194,0.0001902576],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001864029,"about_ca_system_score_gemma":0.002711512,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.02038823,"about_ca_topic_score_gemma":0.01179437,"domain_scores_codex":[0.9989138,0.0005862358,0.00004435536,0.0001803917,0.0001560566,0.0001192305],"domain_scores_gemma":[0.9802941,0.01738484,0.0005193609,0.0007478312,0.0006947265,0.0003591357],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00005618792,0.00003247686,0.0005375958,0.000007745045,0.00001515572,0.00001711956,0.00002197543,0.9955755,0.0001173345,0.001900668,0.00007487083,0.001643273],"study_design_scores_gemma":[0.000003884661,0.000006725144,0.00004200831,0.000001029707,0.000002207085,0.000002904251,0.000001856301,0.9992482,0.00003940569,0.000637684,0.00001281474,0.00000124602],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3474365,0.0004575127,0.6476871,0.001122417,0.00007490806,0.0001510177,0.0001425505,0.0006001934,0.002327828],"genre_scores_gemma":[0.9521127,0.00008963105,0.04642723,0.000110144,0.00002956824,0.0001061891,0.000128033,0.00007516187,0.0009213194],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.02038823,"threshold_uncertainty_score":0.04053915,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0656222157227938,"score_gpt":0.3315865052443263,"score_spread":0.2659642895215325,"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."}}