Dementia Modeling Using Deep Neural Networks Lead Investigator: Mostafa Mehdipour Ghazi Institution : University of Copenhagen E-Mail : mehdipour@biomediq.com Proposal ID : 1148 Proposal Description: The project involves the development of data mining and deep machine learning methods to identify progression and stage transitions in different types of dementia such as the Alzheimer's disease (AD) and vascular dementia (VaD) by introducing parametric and non-parametric curves for modeling imaging biomarkers and clinical data. In this project, it is hypothesized that the state-of-the-art parametric and regression-based methods in disease progression modeling of AD can also model VaD and mixed dementia (MixD) progression. On the other hand, non-parametric methods, such as recurrent neural networks, can improve dementia modeling performance applied to longitudinal data beyond parametric methods. Furthermore, imaging biomarkers of vascular pathologies are valuable to be studied in AD, and sufficient for modeling of VaD.