{"id":"W3185951817","doi":"10.2196/20492","title":"Assessing the Performance of Clinical Natural Language Processing Systems: Development of an Evaluation Methodology","year":2021,"lang":"en","type":"article","venue":"JMIR Medical Informatics","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":64,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Computer science; Robustness (evolution); Annotation; Representativeness heuristic; Task (project management); Population; Software; Data collection; Health informatics; Artificial intelligence; Natural language processing; Data science; Data mining; Public health; Medicine","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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01103032,0.00009212481,0.00033452,0.00005798463,0.0001193717,0.00007217352,0.0007170708,0.0001503184,0.00001552668],"category_scores_gemma":[0.00225986,0.00005985977,0.00004091657,0.0003625313,0.0001114112,0.0007451322,0.0002734496,0.0005906514,0.000002640986],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003602376,"about_ca_system_score_gemma":0.002381537,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000007558237,"about_ca_topic_score_gemma":0.000006135321,"domain_scores_codex":[0.9954573,0.001217831,0.001619678,0.0001164155,0.001396359,0.0001924032],"domain_scores_gemma":[0.9972765,0.0008059821,0.0008016616,0.0004259885,0.0005813661,0.0001084324],"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":[0.000002721116,0.00004238752,0.002742244,0.001107622,0.00001205187,0.000002145573,0.02764036,0.0002911588,0.00001713836,0.0002456555,0.00001117691,0.9678853],"study_design_scores_gemma":[0.0002406683,0.00004458985,0.0184082,0.0003609291,0.000006581779,0.00005228841,0.005745898,0.9744197,0.0004412053,0.000006652523,0.0002067738,0.0000664969],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9229773,0.0005243165,0.07524707,0.0002096758,0.0005077253,0.0002025721,2.463953e-7,0.0000378033,0.0002932576],"genre_scores_gemma":[0.7588082,0.000004138466,0.2408809,0.0002064894,0.00006165121,0.00001736154,0.00001006025,0.000003423626,0.000007807216],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9741285,"threshold_uncertainty_score":0.4224744,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1655180781338516,"score_gpt":0.5212661231110078,"score_spread":0.3557480449771562,"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."}}