{"id":"W3110818709","doi":"10.1016/j.nicl.2021.102712","title":"Cross-cohort generalizability of deep and conventional machine learning for MRI-based diagnosis and prediction of Alzheimer’s disease","year":2021,"lang":"en","type":"preprint","venue":"NeuroImage Clinical","topic":"Dementia and Cognitive Impairment Research","field":"Medicine","cited_by":4,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"National Institute on Aging; National Institute of Biomedical Imaging and Bioengineering; Canadian Institutes of Health Research; Horizon 2020 Framework Programme; Genentech; National Institutes of Health; Medical Delta; IXICO; H. Lundbeck A/S; Servier; Eisai; Hartstichting; Northern California Institute for Research and Education; Pfizer; Biogen; BioClinica; Janssen Research and Development; Nvidia; F. Hoffmann-La Roche; University of Southern California; Novartis Pharmaceuticals Corporation; U.S. Department of Defense; Eli Lilly and Company; Bristol-Myers Squibb; European Commission; GE Healthcare; Alzheimer's Disease Neuroimaging Initiative; Johnson and Johnson; Meso Scale Diagnostics; Merck; Horizon 2020; Fujirebio Europe; Alzheimer's Association; Foundation for the National Institutes of Health","keywords":"Generalizability theory; Support vector machine; Artificial intelligence; Dementia; Convolutional neural network; Cohort; Alzheimer's Disease Neuroimaging Initiative; Machine learning; Cross-validation; Pattern recognition (psychology); Computer science; Neuroimaging; Medicine; Disease; Internal medicine; Mathematics; Statistics; Psychiatry","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":[0.05100074,0.0008815469,0.0007037152,0.001604069,0.0006164869,0.001918888,0.0009744792,0.001069065,0.001422591],"category_scores_gemma":[0.06718647,0.0004532,0.001485366,0.0005426776,0.001238794,0.001282594,0.002444539,0.001592399,0.0006314875],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0005366193,"about_ca_system_score_gemma":0.0004808719,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002982178,"about_ca_topic_score_gemma":0.003361581,"domain_scores_codex":[0.9807554,0.01075368,0.001161905,0.004900773,0.001973581,0.0004546778],"domain_scores_gemma":[0.9400871,0.03220993,0.003361302,0.01846642,0.004872883,0.001002247],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.003410414,0.0003330263,0.9397916,0.0001167808,0.006414722,0.0002417181,0.0009647029,0.006931567,0.005395353,0.0008694821,0.001518358,0.03401217],"study_design_scores_gemma":[0.0001672586,0.002482022,0.9530911,0.00008512206,0.001431327,0.0007568881,0.0004637728,0.02994936,0.005260867,0.002638088,0.003591212,0.00008290308],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9672118,0.00135612,0.02627322,0.0002999014,0.0002795332,0.0002126617,0.001450501,0.0002149673,0.00270135],"genre_scores_gemma":[0.9956981,0.0001305158,0.001949381,0.0001355523,0.00006337521,0.00006017728,0.001487033,0.00006704872,0.0004089289],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.05100074,"threshold_uncertainty_score":0.2697209,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1017986897401413,"score_gpt":0.4314202032894592,"score_spread":0.3296215135493179,"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."}}