Modeling Neurodegenerative Disease Progression Towards Causal Interference with Longitudinal Data Lead Investigator: Alexander Lavin Institution : Latent Sciences E-Mail : lavin@latentsci.com Proposal ID : 1299 Proposal Description: We are developing advanced machine learning (ML) methods for predicting Alzheimer's disease progression. We look to model longitudinal fluid biomarkers, to predict both biomarker- and neurodegeneration- trajectories for a given individual. Our aim is to provide personalized, precision modeling of an individual's neurodegen processes, crucially before they show cognitive and behavioral symptoms. The resulting ML system can be used for applications such as patient stratification and companion diagnostic in clinical trials, precise prognosis for patient care, and more. The objectives for utilizing NACC data are to further validate our ML methods (beyond the use of ADNI). We specifically wish to use the data at the moment to develop causal reasoning algorithms that can move beyond correlative analyses of biomarkers. That is, we seek to derive from the data possible causal links between fluid biomarkers, as well as other neurodegen biomarkers such as cognitive assessments, and also precursor APOE4. This study is a collaboration between Latent Sciences, a software company developing disease modeling systems, and the Turing Institute. The respective PI's are Alexander Lavin (lavin@latentsci.com) and Mihaela van der Schaar. Please reach out with specific questions/comments.