{"id":"W4404667492","doi":"10.2196/60095","title":"Developing an ICD-10 Coding Assistant: Pilot Study Using RoBERTa and GPT-4 for Term Extraction and Description-Based Code Selection","year":2024,"lang":"en","type":"article","venue":"JMIR Formative Research","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":10,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Preprint; Coding (social sciences); Selection (genetic algorithm); Code (set theory); Term (time); Computer science; Statistics; Mathematics; Artificial intelligence; World Wide Web; Programming language","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":[0.0009361859,0.0001088223,0.0001107486,0.0001653717,0.0004255942,0.0002181279,0.00006868722,0.00008846608,0.000005731992],"category_scores_gemma":[0.0001180802,0.00009253349,0.00001749366,0.0001909875,0.0001535241,0.00003174559,0.00006040711,0.0001835675,0.000001022837],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00008383029,"about_ca_system_score_gemma":0.0001526727,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003926562,"about_ca_topic_score_gemma":0.0001932092,"domain_scores_codex":[0.9988878,0.0001709855,0.0001703887,0.0003073245,0.0001894426,0.0002740989],"domain_scores_gemma":[0.9995144,0.0001260744,0.00003183408,0.00008837636,0.0001631872,0.00007614148],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"bench_or_experimental","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001038434,0.0004916271,0.006042874,0.0007672195,0.0001585451,0.000007234969,0.003467804,0.00001716828,0.8981954,0.0003678544,0.0009981627,0.08844762],"study_design_scores_gemma":[0.007764569,0.05600379,0.1991202,0.00177077,0.0001665053,0.0002659252,0.03009528,0.3677284,0.2697702,0.001762571,0.06339134,0.002160443],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9374069,0.0003911296,0.06146749,0.0001225429,0.00007527431,0.0004446348,0.00001661933,0.00003155921,0.00004383438],"genre_scores_gemma":[0.9963333,0.00003978124,0.003213304,0.00001512589,0.00009407283,0.0001255601,0.0000450062,0.0000151815,0.0001186883],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6284252,"threshold_uncertainty_score":0.3773406,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1830089616067131,"score_gpt":0.4688349562210808,"score_spread":0.2858259946143677,"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."}}