Application of Big Data Analytical Approaches to NACC UDS data

 
   Lead Investigator:    Hiroko Dodge
   Institution      :    Oregon Health  Science University
   E-Mail           :    dodgeh@ohsu.edu
   Proposal ID      :    536


   Publications:


1  Lin M, Gong P, Yang T, Ye J, Albin R, Dodge H
   Big Data Analytical Approaches to the NACC Dataset: Aiding Preclinical Trial Enrichment
   Journal Article  Alzheimer Dis Assoc Disord,32:18-27,2018,PMC5854492

   PUBLISHED





   Proposal Description:
   (1) Identifying those at risk of developing MCI and Alzheimer???s disease during the pre-symptomatic phase is becoming increasingly important for cost-effective clinical trials and also for deriving maximum benefit from currently available treatment strategies.  Although biomarkers have been extensively examined as early indicators of the pathological process for AD, assessing these biomarkers is expensive and challenging. In the proposed project, we will apply big data analytical methods to over 300 clinical markers collected in the NACC UDS and derive effective algorithms for identifying those at risk of developing MCI and AD.   (2) We test the power of integrating various baseline variables for predicting the conversion from normal to MCI (and AD) as well as trajectories of cognitive test scores over time. Our first step is to use only clinical data available in UDS.  Depending on data availability and accessibility, we will expand the analysis to include MRI and    GWAS data using the subset of subjects. The analytical models we experiment include, but not limited to, sparse logistic regression with stability selection and an RBF (Gaussian radial basis function) kernel modelling. We will evaluate the prediction performance of different algorithms in terms of the leave-one-out AUC score. This is an exploratory project.