{"id":"W4411019801","doi":"10.1109/jtehm.2025.3576570","title":"Improving Transformer Performance for French Clinical Notes Classification Using Mixture of Experts on a Limited Dataset","year":2025,"lang":"en","type":"article","venue":"IEEE Journal of Translational Engineering in Health and Medicine","topic":"Biomedical Text Mining and Ontologies","field":"Biochemistry, Genetics and Molecular Biology","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"Centre Hospitalier Universitaire Sainte-Justine; Université de Montréal; École de Technologie Supérieure","funders":"Fonds de recherche du Québec – Nature et technologies; Natural Sciences and Engineering Research Council of Canada; Fonds de Recherche du Québec - Santé; Institut de Valorisation des Données","keywords":"Computer science; Transformer; Artificial intelligence; Natural language processing; Pattern recognition (psychology); Machine learning; Engineering","routes":{"ca_aff":true,"ca_fund":true,"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":[0.0005829224,0.00006832011,0.0002074663,0.0001250614,0.0000225009,0.000001462889,0.00005350741,0.0001001892,6.754597e-7],"category_scores_gemma":[0.0003222047,0.00005163281,0.00003295037,0.00007327826,0.00005488987,0.000004255576,0.000001355541,0.0001123885,7.342392e-9],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00000806004,"about_ca_system_score_gemma":0.0001751039,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001159656,"about_ca_topic_score_gemma":0.000003487075,"domain_scores_codex":[0.9991411,0.00001908188,0.0005535042,0.00009162326,0.00009320717,0.0001014707],"domain_scores_gemma":[0.9995,0.0001863484,0.0001311271,0.00005477302,0.00005708095,0.00007063297],"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":[0.002583449,0.0003737173,0.0327349,0.004343935,0.0002225106,0.00000316693,0.0009043817,0.01978423,0.3265884,0.0002050652,0.003687645,0.6085686],"study_design_scores_gemma":[0.03272806,0.02091643,0.2130352,0.01153927,0.0002328079,0.0001730055,0.0006170834,0.5633218,0.03430376,0.00019494,0.1221356,0.000802068],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.8177645,0.004205844,0.1747127,0.002736087,0.0003950068,0.0001389646,0.00004050346,0.000002338089,0.000004030024],"genre_scores_gemma":[0.9895619,0.0008906664,0.008891274,0.0003325187,0.0002321552,0.000002562753,0.00008160163,0.000004291581,0.000003009349],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.6077665,"threshold_uncertainty_score":0.2105525,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0660339813952497,"score_gpt":0.382899591068407,"score_spread":0.3168656096731573,"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."}}