{"id":"W4405623317","doi":"10.2196/67056","title":"Automated Pathologic TN Classification Prediction and Rationale Generation From Lung Cancer Surgical Pathology Reports Using a Large Language Model Fine-Tuned With Chain-of-Thought: Algorithm Development and Validation Study","year":2024,"lang":"en","type":"article","venue":"JMIR Medical Informatics","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":11,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Korea Health Industry Development Institute","keywords":"Context (archaeology); Computer science; Artificial intelligence; Medicine; Lung cancer; Natural language processing; Parsing; Medical physics; Pathology; Machine learning","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":[0.0009640084,0.0001367535,0.0001988495,0.0001311304,0.0001440728,0.0001004754,0.00009668885,0.0001585098,0.000008633623],"category_scores_gemma":[0.00006354517,0.0001052372,0.00001367867,0.0002626793,0.00004286718,0.0004521615,0.000120366,0.000251868,4.438131e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00008646891,"about_ca_system_score_gemma":0.0003821983,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000024106,"about_ca_topic_score_gemma":0.00002328217,"domain_scores_codex":[0.9981259,0.0001517645,0.0007061042,0.0002489432,0.0006011263,0.0001661219],"domain_scores_gemma":[0.999229,0.00009326033,0.0002490588,0.0002063599,0.0001139111,0.0001084204],"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":[0.00005913324,0.0008385361,0.1142564,0.001751783,0.0002886408,0.002128587,0.3920617,0.01700079,0.001092467,0.006351496,0.0006602273,0.4635102],"study_design_scores_gemma":[0.0003750091,0.00007120922,0.008852425,0.0001204234,0.00001763343,0.0002379605,0.0005566372,0.9894994,0.00007597632,0.00003626178,0.00005234666,0.0001046439],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6253945,0.0001919484,0.3734862,0.0001676836,0.0001105104,0.0004210337,0.00001304086,0.0002074993,0.000007611521],"genre_scores_gemma":[0.8589921,0.00001895441,0.1404629,0.00005959582,0.0000838921,0.0001582352,0.0002012756,0.000008638686,0.0000144279],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.9724987,"threshold_uncertainty_score":0.4291447,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0325633751547172,"score_gpt":0.3454929506506829,"score_spread":0.3129295754959657,"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."}}