{"id":"W4283516214","doi":"10.1016/j.jval.2022.04.1025","title":"HSD23 Using Natural Language Processing of Unstructured EMR Data to Describe Canadian Patients with ASCVD","year":2022,"lang":"en","type":"article","venue":"Value in Health","topic":"Aortic Thrombus and Embolism","field":"Medicine","cited_by":0,"is_retracted":false,"has_abstract":false,"ca_institutions":"University of British Columbia; St. Michael's Hospital; University of Regina; Novartis (Canada)","funders":"","keywords":"Electronic medical record; Computer science; Medical record; Natural language processing; Epidemiology; Artificial intelligence; Data science; Medicine; Internal medicine; Internet privacy","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":true,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.000278168,0.00009123026,0.0002617698,0.0001964658,0.0001461407,0.000006991513,0.0001692705,0.00002230607,0.00005357816],"category_scores_gemma":[0.00006448026,0.00007712604,0.00001177157,0.0004494711,0.00002370372,0.00006534026,0.0001004837,0.000242865,5.94398e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004795544,"about_ca_system_score_gemma":0.001754383,"about_ca_topic_candidate":true,"about_ca_topic_consensus":true,"about_ca_topic_score_codex":0.4228607,"about_ca_topic_score_gemma":0.03136843,"domain_scores_codex":[0.9987552,0.00006748297,0.0002737819,0.0002400043,0.0003106129,0.0003529552],"domain_scores_gemma":[0.9992177,0.00001192008,0.0001003727,0.0003699312,0.00004265921,0.0002574007],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"observational","study_design_scores_codex":[0.0008043497,0.0004425228,0.7800195,0.00222772,0.00009496357,0.000237382,0.06121691,0.01882309,0.0005918929,0.0001999148,0.00277168,0.1325701],"study_design_scores_gemma":[0.003280882,0.0007336314,0.9273512,0.0009050837,0.00005512237,0.00008656817,0.007002533,0.05736129,0.00004983752,0.00003314258,0.002804446,0.0003363102],"study_design_candidate":"observational","study_design_consensus":"observational","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9974979,0.0007570927,0.00001913558,0.0008328058,0.0001870795,0.0004702843,0.000172955,0.00001264156,0.00005008375],"genre_scores_gemma":[0.9935089,0.000001681183,0.004804499,0.001382924,0.00004760196,0.000003394438,0.0002085131,0.00002113851,0.00002130089],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.3914923,"threshold_uncertainty_score":0.9863065,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06000136648612014,"score_gpt":0.3211107076274071,"score_spread":0.2611093411412869,"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."}}