{"id":"W4407037890","doi":"10.2196/68863","title":"Performance Improvement of a Natural Language Processing Tool for Extracting Patient Narratives Related to Medical States From Japanese Pharmaceutical Care Records by Increasing the Amount of Training Data: Natural Language Processing Analysis and Validation Study","year":2025,"lang":"en","type":"article","venue":"JMIR Medical Informatics","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"","keywords":"Computer science; Natural language processing; Natural (archaeology); Natural language; Narrative; Artificial intelligence; Linguistics; History","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":[0.001970652,0.0002102217,0.0004562461,0.0002200559,0.0002764956,0.0001443663,0.0009203356,0.0001188505,0.00000993596],"category_scores_gemma":[0.00195481,0.0001440971,0.00005050914,0.001023073,0.0001250446,0.000868253,0.0006414258,0.0007428402,1.731005e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00008789435,"about_ca_system_score_gemma":0.0004550776,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0003272276,"about_ca_topic_score_gemma":0.00006352085,"domain_scores_codex":[0.9966317,0.0002213667,0.00127169,0.0002995711,0.001253324,0.0003223542],"domain_scores_gemma":[0.9976689,0.0008482985,0.0005964041,0.0004320924,0.0002832189,0.0001711],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00005481881,0.00006068984,0.006877557,0.0007538344,0.000102263,0.000001824018,0.4083074,0.00003814518,0.000185427,0.000001817986,0.00001176352,0.5836045],"study_design_scores_gemma":[0.0005958566,0.0001773882,0.003302603,0.0005399724,0.00008355801,0.000003869634,0.2322059,0.762455,0.0004770538,0.000002101704,0.00002615951,0.0001305156],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9904459,0.00133626,0.006246047,0.0006574034,0.0001429305,0.001046467,0.00003881243,0.00006468641,0.00002150627],"genre_scores_gemma":[0.9891386,0.000009180105,0.01002826,0.0004136077,0.00002679169,0.00008080317,0.0002873895,0.000008916286,0.000006447467],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.7624168,"threshold_uncertainty_score":0.5876109,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01543985965683919,"score_gpt":0.3759784832091741,"score_spread":0.3605386235523349,"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."}}