{"id":"W3174643149","doi":"10.1101/2021.06.25.449906","title":"Towards the Interpretability of Deep Learning Models for Multi-modal Neuroimaging: Finding Structural Changes of the Ageing Brain","year":2021,"lang":"en","type":"preprint","venue":"bioRxiv (Cold Spring Harbor Laboratory)","topic":"Functional Brain Connectivity Studies","field":"Neuroscience","cited_by":4,"is_retracted":false,"has_abstract":true,"ca_institutions":"Artificial Intelligence in Medicine (Canada)","funders":"European Regional Development Fund; Deutsche Forschungsgemeinschaft; Freistaat Sachsen; Bundesministerium für Bildung und Forschung; Institute for Information and Communications Technology Promotion; Korea University; European Commission","keywords":"Neuroimaging; Interpretability; Aging brain; Magnetic resonance imaging; Brain aging; White matter; Convolutional neural network; Deep learning; Psychology; Population; Neuroscience; Artificial intelligence; Medicine; Cognition; Computer science; Radiology","routes":{"ca_aff":true,"ca_fund":false,"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":[0.005057245,0.0009435657,0.0006209896,0.001056572,0.0001990988,0.001513934,0.0008071435,0.001161372,0.0007630166],"category_scores_gemma":[0.01832367,0.0004869505,0.0008833333,0.0004582257,0.0007592862,0.001390741,0.001234463,0.00212152,0.0002771782],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006934465,"about_ca_system_score_gemma":0.0006080742,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.003408923,"about_ca_topic_score_gemma":0.002664052,"domain_scores_codex":[0.9991964,0.0004688347,0.00003113514,0.0001935687,0.00006117675,0.00004889667],"domain_scores_gemma":[0.9948908,0.003420344,0.0005165199,0.0006812002,0.0003517915,0.0001393997],"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":[0.0005163521,0.0002013134,0.0550376,0.0001973581,0.0007335278,0.0003164994,0.0003090945,0.8028318,0.01836538,0.008787039,0.002870919,0.1098332],"study_design_scores_gemma":[0.000005156589,0.00002151037,0.002970136,0.00001336449,0.00001508819,0.0000217111,0.00001376287,0.9870654,0.0009587917,0.008769555,0.0001385554,0.000006902345],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3831236,0.0008334618,0.6114197,0.001623712,0.0000872189,0.00004434123,0.0007628364,0.00114172,0.0009634178],"genre_scores_gemma":[0.9719064,0.0001450538,0.02687531,0.0001201068,0.00004519206,0.00002692259,0.000345256,0.00006582635,0.0004698754],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005057245,"threshold_uncertainty_score":0.02674562,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.06837872270315609,"score_gpt":0.2740993439355229,"score_spread":0.2057206212323668,"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."}}