{"id":"W4414615361","doi":"10.7717/peerj-cs.3208","title":"Transformers and capsule networks <i>vs</i> classical ML on clinical data for alzheimer classification","year":2025,"lang":"en","type":"article","venue":"PeerJ Computer Science","topic":"Machine Learning in Healthcare","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"National Institute of Biomedical Imaging and Bioengineering; Canadian Institutes of Health Research; GE Healthcare; Agenția Națională pentru Cercetare și Dezvoltare; IXICO; National Institutes of Health; H. Lundbeck A/S; Servier; National Institute on Aging; Alzheimer's Association; Universidad de Caldas; Biogen; Roche; Eli Lilly and Company; Novartis Pharmaceuticals Corporation","keywords":"Interpretability; Oversampling; Dementia; Convolutional neural network; Deep learning; Feature selection; Boosting (machine learning); Cognition; Neuroimaging","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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.003659926,0.0002110822,0.0003078295,0.0002168152,0.0006229138,0.0005177967,0.003608088,0.0001277051,0.000001520669],"category_scores_gemma":[0.0002459993,0.00018698,0.00006510962,0.001044294,0.0006142246,0.0008561998,0.0009622168,0.0004953713,0.000007842442],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0000491807,"about_ca_system_score_gemma":0.0004821073,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002414534,"about_ca_topic_score_gemma":0.00001239426,"domain_scores_codex":[0.9964812,0.0001838229,0.0005548359,0.001660787,0.0005198537,0.0005995367],"domain_scores_gemma":[0.9968596,0.0007297511,0.0001348532,0.00176612,0.0002383193,0.0002713811],"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":[0.00002778045,0.0001069555,0.002997612,0.00003000063,0.00002334373,0.000002258449,0.0001234722,0.001401069,0.00002975619,0.0837418,0.01686978,0.8946462],"study_design_scores_gemma":[0.000431905,0.0003050164,0.03708873,0.00004373338,0.00001553001,0.000005809994,0.000003895726,0.9305803,0.00002340136,0.0006608568,0.03066005,0.0001808059],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.002458801,0.0001983491,0.9654433,0.02816241,0.002531691,0.0005327564,0.000009989965,0.0002084736,0.0004542379],"genre_scores_gemma":[0.7559552,0.00006400826,0.2354,0.007949375,0.0004456954,0.00004208169,0.0000298036,0.0000135062,0.0001004239],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9291792,"threshold_uncertainty_score":0.7624824,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1348626890089853,"score_gpt":0.4245521902458566,"score_spread":0.2896895012368713,"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."}}