{"id":"W4403349562","doi":"10.1016/j.bas.2024.103579","title":"Large Language Models Can Extract Tabular Data from Unstructured Patient Records To Be Used in Large-Scale Retrospective Studies","year":2024,"lang":"en","type":"article","venue":"Brain and Spine","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"ca_institutions":"University of Calgary","funders":"","keywords":"Scale (ratio); Unstructured data; Computer science; Natural language processing; Data science; Data mining; Geography; Cartography; Big data","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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004877194,0.0001782496,0.0003349266,0.000155972,0.00009129533,0.0001245343,0.0005320264,0.00006724451,0.00002103447],"category_scores_gemma":[0.000263435,0.0001504169,0.00002882011,0.0004950797,0.00001835701,0.0002807624,0.0008277525,0.0003339283,0.000004165226],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00008274208,"about_ca_system_score_gemma":0.00006537959,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0008644492,"about_ca_topic_score_gemma":0.01127214,"domain_scores_codex":[0.9981968,0.0001544421,0.0002697711,0.0007871863,0.0002487943,0.0003429579],"domain_scores_gemma":[0.9988501,0.0001166834,0.00004318994,0.0008338261,0.00003258271,0.000123589],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"qualitative","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00005330114,0.0003536482,0.0410826,0.001336551,0.0003696682,0.001982692,0.4752083,0.001128354,0.000564776,0.04866937,0.01460153,0.4146492],"study_design_scores_gemma":[0.0007631946,0.0003263684,0.03706197,0.0006652198,0.00001675549,0.00001504076,0.004300717,0.899486,0.00004666315,0.004717337,0.05204397,0.0005567354],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8004797,0.09514382,0.08138986,0.0198126,0.0009076514,0.0007560334,0.001003426,0.0003922106,0.0001147435],"genre_scores_gemma":[0.9728499,0.001083541,0.02389779,0.001686441,0.0001281793,0.00002456523,0.0001495431,0.00002597618,0.0001540518],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.8983577,"threshold_uncertainty_score":0.6290117,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03842732420521019,"score_gpt":0.337877959462927,"score_spread":0.2994506352577169,"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."}}