{"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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001737936,0.001054504,0.0005846745,0.008535706,0.0007017663,0.001738782,0.00115705,0.001432994,0.002290984],"category_scores_gemma":[0.01039512,0.0004460627,0.0009756483,0.002503861,0.0005840314,0.001990677,0.001694654,0.001176796,0.002251017],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005391376,"about_ca_system_score_gemma":0.001739055,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003070675,"about_ca_topic_score_gemma":0.004021891,"domain_scores_codex":[0.9968475,0.0004681444,0.0006358936,0.0009155645,0.0009277572,0.0002052195],"domain_scores_gemma":[0.980877,0.01060801,0.003861922,0.0009879145,0.003054557,0.0006105843],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"observational","study_design_scores_codex":[0.00180449,0.0007560814,0.08651553,0.005118092,0.0003357095,0.009326513,0.003064844,0.004367265,0.2876056,0.005555043,0.03883206,0.5567188],"study_design_scores_gemma":[0.0002723623,0.001058647,0.2155289,0.001417261,0.00137149,0.02673306,0.005203478,0.3357863,0.2768867,0.01551671,0.1197152,0.0005098489],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6436656,0.006724829,0.2662285,0.002440827,0.001101914,0.00120865,0.03969075,0.02976257,0.009176428],"genre_scores_gemma":[0.7038888,0.001345575,0.2553154,0.0003332342,0.0004969155,0.0003539471,0.03426226,0.0004852063,0.003518631],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.008535706,"threshold_uncertainty_score":0.009191215,"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."}}