{"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/neural-processes-mixed-effect-models-for-deep","title":"Neural Processes Mixed-Effect Models for Deep Normative Modeling of Clinical Neuroimaging Data","arxiv_id":"1812.04998","date":"2018-12-12","proceeding":null,"authors":["Seyed Mostafa Kia","Andre F. Marquand"],"abstract":"Normative modeling has recently been introduced as a promising approach for\nmodeling variation of neuroimaging measures across individuals in order to\nderive biomarkers of psychiatric disorders. Current implementations rely on\nGaussian process regression, which provides coherent estimates of uncertainty\nneeded for the method but also suffers from drawbacks including poor scaling to\nlarge datasets and a reliance on fixed parametric kernels. In this paper, we\npropose a deep normative modeling framework based on neural processes (NPs) to\nsolve these problems. To achieve this, we define a stochastic process\nformulation for mixed-effect models and show how NPs can be adopted for\nspatially structured mixed-effect modeling of neuroimaging data. This enables\nus to learn optimal feature representations and covariance structure for the\nrandom-effect and noise via global latent variables. In this scheme, predictive\nuncertainty can be approximated by sampling from the distribution of these\nglobal latent variables. On a publicly available clinical fMRI dataset, we\ncompare the novelty detection performance of multivariate normative models\nestimated by the proposed NP approach to a baseline multi-task Gaussian process\nregression approach and show substantial improvements for certain diagnostic\nproblems.","url_abs":"http://arxiv.org/abs/1812.04998v2","url_pdf":"http://arxiv.org/pdf/1812.04998v2.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":"neural-processes-mixed-effect-models-for-deep","repo_url":"https://github.com/smkia/DNM","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"diagnostic","task_name":"Diagnostic"},{"task_slug":"novelty-detection","task_name":"Novelty Detection"},{"task_slug":"regression-1","task_name":"regression"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1812.04998","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}