{"id":"W4206443342","doi":"10.1007/s00330-021-08467-8","title":"Automatic detection of actionable findings and communication mentions in radiology reports using natural language processing","year":2022,"lang":"en","type":"article","venue":"European Radiology","topic":"Radiology practices and education","field":"Medicine","cited_by":9,"is_retracted":false,"has_abstract":false,"ca_institutions":"","funders":"Newfoundland and Labrador","keywords":"Flagging; Classifier (UML); Artificial intelligence; Test set; Computer science; Receiver operating characteristic; Preprocessor; Machine learning; Documentation; Natural language processing; Medicine; Neuroradiology; Neurology","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":[0.001028581,0.00007821694,0.0002168947,0.0002485336,0.0002352826,0.000005793695,0.0000577213,0.00003472772,0.00007798208],"category_scores_gemma":[0.000212457,0.00008008648,0.00002932804,0.0002177743,0.0001176163,0.0001243118,0.00006754683,0.0003540651,0.000001141599],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001572202,"about_ca_system_score_gemma":0.00006197733,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00008966617,"about_ca_topic_score_gemma":0.000007017173,"domain_scores_codex":[0.9984998,0.0007409218,0.0003500578,0.000199724,0.00006133423,0.0001482359],"domain_scores_gemma":[0.9993577,0.0001031838,0.0002736887,0.0002070132,0.00002861279,0.00002981861],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"observational","study_design_scores_codex":[0.0005026888,0.0005276727,0.1835875,0.0003518622,0.0002084518,0.0004916248,0.02489259,0.000981592,0.6799803,0.0002094253,0.0005340798,0.1077322],"study_design_scores_gemma":[0.001652478,0.0007174675,0.8675061,0.00008416743,0.0002232489,0.04936363,0.01183097,0.06427358,0.0009606455,0.0001330336,0.002992367,0.0002622635],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9934291,0.004832642,0.0001072504,0.0003313336,0.0002684706,0.000205656,9.282268e-7,0.00003446609,0.0007901149],"genre_scores_gemma":[0.9981169,0.00007475012,0.00140837,0.0001175201,0.00004720527,0.00001418888,0.00005099652,0.00001512318,0.0001549855],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6839186,"threshold_uncertainty_score":0.3265832,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01902958951043082,"score_gpt":0.3101502478497598,"score_spread":0.291120658339329,"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."}}