Building Deep Learning based Survival Models for Opportunistic Screening of
AD on MRI
 


   Note: An Official Proposal has not been Submitted to NACC.


   Lead Investigator:    Mark Cicero
   Institution      :    16 Bit Inc.
   E-Mail           :    mark@16bit.ai
   Proposal ID      :    1766



   Proposal Description:
   Deep learning based survival models through the use of convolutional neural networks (CNNs) can be used to extract features from images that enable the prediction that an individual might develop a disease over a specific time period. We hypothesize that the use of 3D MRI data and other clinical covariates can be used to predict one's individual risk of developing AD. We will develop models to do this and test their performance by measuring c-index, brier score and AUC for the classification of AD. We believe survival models able to predict future AD risk would be clinically useful and can be deployed in an opportunistic fashion to screen patients for their risk of developing the disease while undergoing routine CT or MRI scans of the brain for any indication.