{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/spectral-graph-convolutions-for-population","title":"Spectral Graph Convolutions for Population-based Disease Prediction","arxiv_id":"1703.03020","date":"2017-03-08","proceeding":null,"authors":["Sarah Parisot","Sofia Ira Ktena","Enzo Ferrante","Matthew Lee","Ricardo Guerrerro Moreno","Ben Glocker","Daniel Rueckert"],"abstract":"Exploiting the wealth of imaging and non-imaging information for disease\nprediction tasks requires models capable of representing, at the same time,\nindividual features as well as data associations between subjects from\npotentially large populations. Graphs provide a natural framework for such\ntasks, yet previous graph-based approaches focus on pairwise similarities\nwithout modelling the subjects' individual characteristics and features. On the\nother hand, relying solely on subject-specific imaging feature vectors fails to\nmodel the interaction and similarity between subjects, which can reduce\nperformance. In this paper, we introduce the novel concept of Graph\nConvolutional Networks (GCN) for brain analysis in populations, combining\nimaging and non-imaging data. We represent populations as a sparse graph where\nits vertices are associated with image-based feature vectors and the edges\nencode phenotypic information. This structure was used to train a GCN model on\npartially labelled graphs, aiming to infer the classes of unlabelled nodes from\nthe node features and pairwise associations between subjects. We demonstrate\nthe potential of the method on the challenging ADNI and ABIDE databases, as a\nproof of concept of the benefit from integrating contextual information in\nclassification tasks. This has a clear impact on the quality of the\npredictions, leading to 69.5% accuracy for ABIDE (outperforming the current\nstate of the art of 66.8%) and 77% for ADNI for prediction of MCI conversion,\nsignificantly outperforming standard linear classifiers where only individual\nfeatures are considered.","url_abs":"http://arxiv.org/abs/1703.03020v3","url_pdf":"http://arxiv.org/pdf/1703.03020v3.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"spectral-graph-convolutions-for-population","repo_url":"https://github.com/parisots/population-gcn","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"GPL-3.0"}}],"tasks":[{"task_slug":"disease-prediction","task_name":"Disease Prediction"},{"task_slug":"prediction","task_name":"Prediction"}],"methods":[{"method_slug":"gcn","method_name":"GCN"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}