• Epelbaum, Stephane
• Habert, Marie-Odile
• Samper-Gonzalez, Jorge
• Colliot, Olivier
• Burgos, Ninon
• Evgeniou, Theodoros
• Bottani, Simona

Abstract

Various machine learning approaches have been developed for predicting progression to Alzheimer’s disease (AD) in patients with mild cognitive impairment (MCI) from MRI and PET data. Objective comparison of these approaches is nearly impossible because of differences at all steps, fromdata management to image processing and evaluation procedures. Moreover, with a few exceptions, these papers rarely compare their results to that obtained with clinical/cognitive data only, a critical point to demonstrate the practical utility of neuroimaging in this context. We previously proposed a framework for the reproducible evaluation of ML algorithms for AD classification. This framework was applied to AD classification using unimodal neuroimaging data (T1 MRI and FDG PET). Here, we extend our previous workto the combination of multimodal clinical and neuroimaging data for predicting progression to AD among MCI patients.All the code is publicly available at: https://github.com/aramis-lab/AD-ML.

Topics

• impedance spectroscopy
• machine learning