{"id":"W2768609493","doi":"10.1097/ede.0000000000000787","title":"Can We Train Machine Learning Methods to Outperform the High-dimensional Propensity Score Algorithm?","year":2017,"lang":"en","type":"article","venue":"Epidemiology","topic":"Statistical Methods and Inference","field":"Mathematics","cited_by":64,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University Health Centre; McGill University; University of British Columbia; Centre for Advancing Health Outcomes; Jewish General Hospital; Providence Health Care","funders":"Canadian Institutes of Health Research","keywords":"Propensity score matching; Confounding; Machine learning; Covariate; Computer science; Selection bias; Artificial intelligence; Random forest; Algorithm; Elastic net regularization; Feature selection; Statistics; Mathematics","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":[0.03399668,0.001154802,0.001783239,0.001651351,0.0005578657,0.002052939,0.001700089,0.002500431,0.002540714],"category_scores_gemma":[0.1282766,0.0005626233,0.001146485,0.001993837,0.00149882,0.005499413,0.001448366,0.002773556,0.001299817],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009312844,"about_ca_system_score_gemma":0.002395435,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002631247,"about_ca_topic_score_gemma":0.002262734,"domain_scores_codex":[0.9925185,0.005550572,0.0003704675,0.0006534598,0.0007012393,0.0002056618],"domain_scores_gemma":[0.9557698,0.03474466,0.002049551,0.004168422,0.002833428,0.0004341607],"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":[0.0004177673,0.000435254,0.02903739,0.0005175788,0.0009294472,0.00007601428,0.0001889452,0.404718,0.0006457595,0.05456432,0.01277359,0.495696],"study_design_scores_gemma":[0.0001582167,0.0001854743,0.00293347,0.0001485479,0.00006201505,0.00007065731,0.00005294103,0.8937488,0.0007012581,0.09722921,0.004666649,0.00004265371],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.04300119,0.003583404,0.9375479,0.01180653,0.0004447753,0.0001169949,0.0002577859,0.001052545,0.00218881],"genre_scores_gemma":[0.3746721,0.002564758,0.6164848,0.002870713,0.0006681754,0.0003548956,0.0006387663,0.0002021788,0.001543642],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.03399668,"threshold_uncertainty_score":0.1797937,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.3766227571966274,"score_gpt":0.4804717655727531,"score_spread":0.1038490083761257,"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."}}