{"id":"W4390789641","doi":"10.1017/s1351324923000542","title":"Lightweight transformers for clinical natural language processing","year":2024,"lang":"en","type":"article","venue":"Natural Language Engineering","topic":"Topic Modeling","field":"Computer Science","cited_by":30,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"NIHR Imperial Biomedical Research Centre; Instituto de Salud Carlos III; Medical Research Council; National Institutes of Health; Kementerian Kesihatan Malaysia; All-India Institute of Medical Sciences; Horizon 2020 Framework Programme; Prince Charles Hospital Foundation; Foreign, Commonwealth and Development Office; National Institute for Health Research Health Protection Research Unit; University of Oxford; Norges Forskningsråd; Imperial College London; Conselho Nacional de Desenvolvimento Científico e Tecnológico; University of Cape Town; Public Health England; Wellcome Trust; University College Dublin; Sunnybrook Research Institute; Institut National de la Santé et de la Recherche Médicale; Canadian Institutes of Health Research; National Institute for Health and Care Research; Ministero della Salute; European Federation of Pharmaceutical Industries and Associations; European Commission; Bill and Melinda Gates Foundation","keywords":"Computer science; Transformer; Natural language processing; Artificial intelligence; Electrical engineering","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002547574,0.001314109,0.000578846,0.001684292,0.0003493985,0.00183358,0.002084572,0.0009072938,0.01351506],"category_scores_gemma":[0.01545823,0.0007173746,0.001319676,0.001254858,0.00108187,0.006130395,0.003767738,0.00317759,0.008818846],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0015542,"about_ca_system_score_gemma":0.002915381,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003133788,"about_ca_topic_score_gemma":0.00571743,"domain_scores_codex":[0.9976996,0.0007538392,0.0002667524,0.0005486269,0.0006125403,0.000118685],"domain_scores_gemma":[0.9932497,0.003799125,0.000349893,0.001610354,0.0007749989,0.0002160207],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"bench_or_experimental","study_design_scores_codex":[0.001062609,0.0004112517,0.004386563,0.00102061,0.0001779017,0.0005135506,0.0004802328,0.1242555,0.01213633,0.05540857,0.06832554,0.7318212],"study_design_scores_gemma":[0.0001211837,0.0002119466,0.0004786034,0.0001226599,0.00006660144,0.0003604522,0.0001489833,0.8164868,0.01593537,0.1236837,0.04232388,0.00005985633],"study_design_candidate":"bench_or_experimental","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01791121,0.001352246,0.8986432,0.002445353,0.00032328,0.0005248201,0.005984173,0.06904554,0.003770186],"genre_scores_gemma":[0.3717313,0.001273559,0.5995435,0.00144167,0.0002141247,0.0006950929,0.01659666,0.002284962,0.006219259],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.01351506,"threshold_uncertainty_score":0.04521239,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01169029268271533,"score_gpt":0.3042427404396033,"score_spread":0.2925524477568879,"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."}}