{"id":"W3114255343","doi":"10.1186/s13195-021-00924-2","title":"Improving 3D convolutional neural network comprehensibility via interactive visualization of relevance maps: evaluation in Alzheimer’s disease","year":2021,"lang":"en","type":"preprint","venue":"Alzheimer s Research & Therapy","topic":"Dementia and Cognitive Impairment Research","field":"Medicine","cited_by":5,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"National Institute on Aging; National Institute of Biomedical Imaging and Bioengineering; Canadian Institutes of Health Research; National Institutes of Health; Genentech; IXICO; Deutsches Zentrum für Neurodegenerative Erkrankungen; Freie Universität Berlin; H. Lundbeck A/S; Universität Rostock; Servier; Eisai; Deutsche Forschungsgemeinschaft; Northern California Institute for Research and Education; F. Hoffmann-La Roche; University of Southern California; Pfizer; BioClinica; Biogen; Eli Lilly and Company; Bristol-Myers Squibb; Meso Scale Diagnostics; Alzheimer's Disease Neuroimaging Initiative; Novartis Pharmaceuticals Corporation; Alzheimer's Association; Foundation for the National Institutes of Health","keywords":"Convolutional neural network; Relevance (law); Dementia; Atrophy; Discriminative model; Magnetic resonance imaging; Clinical significance; Computer science; Artificial intelligence; Pattern recognition (psychology); Psychology; Disease; Medicine; Pathology; Radiology","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.004253369,0.002323673,0.0008492072,0.00154498,0.0002483051,0.001220795,0.001306286,0.001351971,0.001180167],"category_scores_gemma":[0.01459095,0.0005349403,0.001416969,0.000491731,0.0004133748,0.001024855,0.001124561,0.001024623,0.000224241],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00111643,"about_ca_system_score_gemma":0.000536838,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.008883614,"about_ca_topic_score_gemma":0.006283167,"domain_scores_codex":[0.9991482,0.0003616796,0.00007390668,0.0001874121,0.0001618931,0.00006677835],"domain_scores_gemma":[0.9945162,0.004141701,0.0003170298,0.0003443992,0.0004915904,0.0001890881],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.003601474,0.0007819292,0.031689,0.001010244,0.001319159,0.0009648592,0.0005704422,0.6630705,0.01701714,0.0008453414,0.004722514,0.2744074],"study_design_scores_gemma":[0.00007042071,0.0003529735,0.004303361,0.00003749615,0.0001077375,0.0001764928,0.00002732517,0.9882504,0.005642959,0.0006258231,0.0003795644,0.00002554544],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"methods","genre_scores_codex":[0.8972727,0.004787097,0.08709263,0.0005035992,0.0001975059,0.0002076643,0.001344337,0.007440768,0.001153675],"genre_scores_gemma":[0.9573423,0.0007425103,0.03961688,0.00008673446,0.0000466183,0.0000774954,0.001369323,0.000217724,0.0005005133],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.008883614,"threshold_uncertainty_score":0.02249426,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1401646399574572,"score_gpt":0.4485761138401771,"score_spread":0.3084114738827199,"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."}}