eprintid: 10073011 rev_number: 20 eprint_status: archive userid: 608 dir: disk0/10/07/30/11 datestamp: 2019-05-02 10:02:58 lastmod: 2021-09-28 22:27:20 status_changed: 2019-05-02 10:02:58 type: article metadata_visibility: show creators_name: Donini, M creators_name: Monteiro, JM creators_name: Pontil, M creators_name: Hahn, T creators_name: Fallgatter, AJ creators_name: Shawe-Taylor, J creators_name: Mourão-Miranda, J creators_name: Alzheimer's Disease Neuroimaging Initiative, . title: Combining heterogeneous data sources for neuroimaging based diagnosis: re-weighting and selecting what is important ispublished: pub divisions: UCL divisions: B04 divisions: C05 divisions: F48 keywords: Feature selection, Multiple kernel learning, Neuroimaging note: © 2019 The Author(s). This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). abstract: Combining neuroimaging and clinical information for diagnosis, as for example behavioral tasks and genetics characteristics, is potentially beneficial but presents challenges in terms of finding the best data representation for the different sources of information. Their simple combination usually does not provide an improvement if compared with using the best source alone. In this paper, we proposed a framework based on a recent multiple kernel learning algorithm called EasyMKL and we investigated the benefits of this approach for diagnosing two different mental health diseases. The well known Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset tackling the Alzheimer Disease (AD) patients versus healthy controls classification task, and a second dataset tackling the task of classifying an heterogeneous group of depressed patients versus healthy controls. We used EasyMKL to combine a huge amount of basic kernels alongside a feature selection methodology, pursuing an optimal and sparse solution to facilitate interpretability. Our results show that the proposed approach, called EasyMKLFS, outperforms baselines (e.g. SVM and SimpleMKL), state-of-the-art random forests (RF) and feature selection (FS) methods. date: 2019-07-15 date_type: published official_url: https://doi.org/10.1016/j.neuroimage.2019.01.053 oa_status: green full_text_type: pub language: eng primo: open primo_central: open_green article_type_text: Journal Article verified: verified_manual elements_id: 1642375 doi: 10.1016/j.neuroimage.2019.01.053 pii: S1053-8119(19)30049-7 lyricists_name: Mourao-Miranda, Janaina lyricists_name: Pontil, Massimiliano lyricists_name: Shawe-Taylor, John lyricists_id: JMOUR63 lyricists_id: MPONT27 lyricists_id: JSHAW87 actors_name: Flynn, Bernadette actors_id: BFFLY94 actors_role: owner full_text_status: public publication: Neuroimage volume: 195 pagerange: 215-231 event_location: United States issn: 1095-9572 citation: Donini, M; Monteiro, JM; Pontil, M; Hahn, T; Fallgatter, AJ; Shawe-Taylor, J; Mourão-Miranda, J; Donini, M; Monteiro, JM; Pontil, M; Hahn, T; Fallgatter, AJ; Shawe-Taylor, J; Mourão-Miranda, J; Alzheimer's Disease Neuroimaging Initiative, .; - view fewer <#> (2019) Combining heterogeneous data sources for neuroimaging based diagnosis: re-weighting and selecting what is important. Neuroimage , 195 pp. 215-231. 10.1016/j.neuroimage.2019.01.053 <https://doi.org/10.1016/j.neuroimage.2019.01.053>. Green open access document_url: https://discovery.ucl.ac.uk/id/eprint/10073011/1/1-s2.0-S1053811919300497-main.pdf