{"id":"W4214586617","doi":"10.2139/ssrn.4004925","title":"Rule-Based Natural Language Processing for Automation of Stroke Data Extraction: A Validation Study","year":2022,"lang":"en","type":"article","venue":"SSRN Electronic Journal","topic":"Acute Ischemic Stroke Management","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":false,"ca_institutions":"University of Toronto; Ottawa Hospital; University of Ottawa","funders":"","keywords":"Natural language processing; Computer science; Automation; Stroke (engine); Artificial intelligence; Data extraction; Extraction (chemistry); Data mining; Data science; MEDLINE; Engineering","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.01322864,0.001040347,0.0007837039,0.002002487,0.0007091796,0.001508834,0.001614634,0.0009436462,0.002928587],"category_scores_gemma":[0.03565762,0.0004279289,0.001419681,0.001140786,0.0007007369,0.001512479,0.001519092,0.0009847705,0.001586707],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006276871,"about_ca_system_score_gemma":0.002070934,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005354802,"about_ca_topic_score_gemma":0.00385482,"domain_scores_codex":[0.9921501,0.003921764,0.001308001,0.001239223,0.001128024,0.0002529813],"domain_scores_gemma":[0.941226,0.04464079,0.001195806,0.005134123,0.007423149,0.000380094],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.01105759,0.01248938,0.1067569,0.004148672,0.002131245,0.002888571,0.005954768,0.02845429,0.07204895,0.001467148,0.007764541,0.7448378],"study_design_scores_gemma":[0.00581604,0.02205665,0.2532392,0.001102496,0.004987532,0.008552587,0.005663814,0.4275985,0.2220386,0.005438101,0.04301211,0.0004942623],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9002696,0.001253733,0.08288194,0.0002379511,0.0001002824,0.002903467,0.006275258,0.002898389,0.003179444],"genre_scores_gemma":[0.833286,0.0006270426,0.1477445,0.0002203794,0.00004319429,0.001358657,0.01446694,0.0002529047,0.002000354],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01322864,"threshold_uncertainty_score":0.06996059,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02166137684476317,"score_gpt":0.3327618095563395,"score_spread":0.3111004327115764,"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."}}