{"id":"W4408406005","doi":"10.1093/jamia/ocaf042","title":"Optimizing the efficiency and effectiveness of data quality assurance in a multicenter clinical dataset","year":2025,"lang":"en","type":"article","venue":"Journal of the American Medical Informatics Association","topic":"Electronic Health Records Systems","field":"Health Professions","cited_by":10,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto; St. Michael's Hospital","funders":"Canada Research Chairs","keywords":"False positive paradox; Data quality; Computer science; Quality assurance; Data collection; Data mining; Chart; Identifier; Medicine; Database; Machine learning; Metric (unit); Statistics; Operations management","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":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch"],"consensus_categories":["metaresearch"],"category_scores_codex":[0.05204526,0.00007989322,0.00061492,0.00007735804,0.0001813413,0.000009378921,0.0009026831,0.0001175112,0.00000358915],"category_scores_gemma":[0.0275404,0.00004237056,0.00005695507,0.0004829142,0.0001630295,0.000239865,0.0003989973,0.001569556,0.000001988991],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005001603,"about_ca_system_score_gemma":0.001578432,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0005447697,"about_ca_topic_score_gemma":0.0003006646,"domain_scores_codex":[0.9898688,0.006182323,0.002749676,0.00007377866,0.0008285877,0.0002968122],"domain_scores_gemma":[0.9741564,0.01989754,0.005178892,0.0004205251,0.0002558981,0.00009077114],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0002355277,0.0001485981,0.9771463,0.001123212,0.0001339649,4.737059e-7,0.001914596,0.00004704571,0.0000136146,0.0001542806,0.0090083,0.01007407],"study_design_scores_gemma":[0.003094513,0.0001614808,0.9358638,0.003499556,0.00007677849,0.000002623023,0.004683537,0.04452769,0.000007201812,0.0001426914,0.007851795,0.00008836242],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9824054,0.0001800324,0.00965185,0.005967092,0.000867485,0.000638413,0.0001303711,0.000004203715,0.0001551112],"genre_scores_gemma":[0.9956263,0.0005377314,0.0008603702,0.002828815,0.000105707,0.000007507685,0.00001673118,0.000004562037,0.00001232776],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.04448064,"threshold_uncertainty_score":0.980651,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.07893946645619035,"score_gpt":0.5384986410175016,"score_spread":0.4595591745613112,"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."}}