{"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":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.139791,0.002051014,0.001254399,0.005474267,0.001120114,0.003300202,0.002578681,0.002407286,0.001057723],"category_scores_gemma":[0.2292007,0.0006674916,0.001777241,0.003808626,0.002149287,0.003021278,0.003185047,0.001230117,0.0003772285],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004372754,"about_ca_system_score_gemma":0.004682856,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001899586,"about_ca_topic_score_gemma":0.001647565,"domain_scores_codex":[0.8190783,0.1243794,0.02272423,0.007359063,0.02518356,0.001275472],"domain_scores_gemma":[0.730989,0.1855527,0.01965166,0.008788748,0.0532953,0.001722468],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.004797071,0.005680761,0.159351,0.005876794,0.002168243,0.000432725,0.00432711,0.07060562,0.02389665,0.008956328,0.004404639,0.709503],"study_design_scores_gemma":[0.002466423,0.03547401,0.1676999,0.001758766,0.002214474,0.0009896222,0.004698789,0.6782603,0.07684676,0.01260629,0.01646049,0.0005241795],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2578703,0.001385403,0.7016975,0.0007173197,0.0001456247,0.03138768,0.00146379,0.0009320157,0.004400347],"genre_scores_gemma":[0.3756862,0.0002812521,0.5963591,0.0001610105,0.00005742084,0.02564533,0.001318633,0.00007570138,0.0004153151],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.860209,"threshold_uncertainty_score":0.7392945,"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."}}