{"id":"W4285104162","doi":"10.2196/38155","title":"Multicenter Validation of Natural Language Processing Algorithms for the Detection of Common Data Elements in Operative Notes for Total Hip Arthroplasty: Algorithm Development and Validation","year":2022,"lang":"en","type":"article","venue":"JMIR Medical Informatics","topic":"Digital Imaging in Medicine","field":"Medicine","cited_by":8,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"National Institutes of Health; National Institute of Arthritis and Musculoskeletal and Skin Diseases; University of Michigan","keywords":"Algorithm; Computer science; Usability; Software portability; Artificial intelligence; Machine learning; Natural language processing; Data mining; Information retrieval","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.1474786,0.001463927,0.0007732703,0.002411045,0.00124554,0.002342643,0.002993818,0.002035403,0.0009329599],"category_scores_gemma":[0.2450214,0.0007583798,0.001962586,0.00142275,0.002535506,0.001661778,0.00406883,0.002344906,0.0007694433],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002322425,"about_ca_system_score_gemma":0.005092343,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004206001,"about_ca_topic_score_gemma":0.003966887,"domain_scores_codex":[0.912428,0.05536408,0.01084788,0.01082189,0.009544769,0.0009934155],"domain_scores_gemma":[0.6386364,0.2296196,0.01885401,0.03645498,0.07402989,0.002405183],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.008408382,0.01025953,0.5095955,0.003195951,0.003376149,0.0007702088,0.0195484,0.0493152,0.04609979,0.002697301,0.007577072,0.3391565],"study_design_scores_gemma":[0.004590203,0.01904368,0.5330552,0.001178503,0.002679042,0.001728451,0.005665102,0.2812202,0.122503,0.003774523,0.02417432,0.0003878641],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.870764,0.0006628502,0.1167526,0.0003395932,0.0001221713,0.006895167,0.001366144,0.001654348,0.001443163],"genre_scores_gemma":[0.7838908,0.000195114,0.2003329,0.0003073294,0.00003905184,0.007363451,0.006831878,0.0004185695,0.0006209826],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1474786,"threshold_uncertainty_score":0.7799507,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03297149042407482,"score_gpt":0.3617818375113848,"score_spread":0.32881034708731,"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."}}