{"id":"W4408776600","doi":"10.1016/j.labinv.2024.102237","title":"15 Identifying the Optimal Architecture for Using Local Large Language Models (LLM) In Autopsy Pathology Reports","year":2025,"lang":"en","type":"article","venue":"Laboratory Investigation","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":false,"ca_institutions":"Institute for Research in Immunology and Cancer; Centre Hospitalier de l’Université de Montréal; Université de Montréal","funders":"","keywords":"Autopsy; Pathology; Architecture; Computer science; Medicine; Natural language processing; Geography; Archaeology","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":["metaresearch"],"consensus_categories":[],"category_scores_codex":[0.002355557,0.0009412801,0.0006290541,0.0009908873,0.0006246316,0.001462443,0.001152238,0.001662521,0.002783351],"category_scores_gemma":[0.006865706,0.0006227799,0.0009688298,0.0004831697,0.0004017666,0.001250436,0.0009992007,0.00179724,0.00192288],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00121762,"about_ca_system_score_gemma":0.002470787,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0103138,"about_ca_topic_score_gemma":0.01601145,"domain_scores_codex":[0.9992693,0.000324359,0.00005520404,0.0001716967,0.00006586619,0.0001136414],"domain_scores_gemma":[0.9978105,0.001326197,0.0001463921,0.0001813579,0.0004320144,0.0001035179],"domain_codex":null,"domain_gemma":"methods","domain_candidate":"methods","domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.001000136,0.0006360626,0.03611127,0.0003306986,0.0003825595,0.0004865085,0.0002866698,0.3400828,0.02297442,0.004578553,0.01164372,0.5814866],"study_design_scores_gemma":[0.0000189337,0.0000950837,0.002225823,0.00002761136,0.00005681385,0.00007229332,0.00007179246,0.9880931,0.005538488,0.003245651,0.0005410917,0.00001324393],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.4096634,0.001601108,0.575277,0.002466279,0.0001779883,0.0002059044,0.001346199,0.005946509,0.003315624],"genre_scores_gemma":[0.8838994,0.0003516178,0.1098764,0.000267788,0.00007281942,0.0001390746,0.001862614,0.0001920752,0.003338307],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9976444,"threshold_uncertainty_score":0.02050757,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02274256757964648,"score_gpt":0.3179683781641615,"score_spread":0.295225810584515,"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."}}