Joe Shaffer, PhD
Joe Shaffer, PhD
Assistant Professor, College of Biosciences
Kansas City University
Dr. Shaffer is an Assistant Professor in the College of Biosciences at Kansas City University (KCU). He earned his PhD in Neurobiology from the University of North Carolina at Chapel Hill and completed his post-doctoral training in the Magnetic Resonance Research Facility at the University of Iowa. Dr. Shaffer’s research focuses on functional neuroimaging in Bipolar Disorder and Schizophrenia. In particular, Dr. Shaffer is interested in the use of machine learning techniques for integrating multi-modal imaging data. At KCU, Dr. Shaffer teaches computer programming and bioinformatics algorithms and is working to launch a new training program in bioinformatics.
Improving Interpretability Using Randomized Input Sampling (Rise)
Human brain imaging generates large quantities of multimodal data that can be challenging to analyze using traditional statistical approaches, particularly given the typical sample size relative to the sheer number of features that can be calculated. Machine learning methods have become an important tool in evaluating these datasets, however, these methods can often be difficult to interpret due to their “black box” nature. Therefore developing methods of feature reduction that improve the interpretability of research results is an important need.
Here we evaluate Randomized Input Sampling for Explanation (RISE) as a method for evaluating which features provide the strongest predictive value. We evaluate its use on a large dataset of structural (T1) and resting-state (BOLD) data downloaded from the NIH National Data Archive in order to measure its efficacy as a feature reduction method for brain imaging data.