Add to Google Calendar
Date:  Thu, May 02, 2019
Time:  1:00pm
Location:  Holmes 389
Speaker:  Dr. John Shepherd

Abstract:

The goal of the AI PHI is to reduce the burden of cancer by utilization of artificial intelligence and deep learning to create diagnostic and risk models using data from Hawaii. We are different that other and regions of the US in terms of our ethnic make up, lifestyle, exposures. Using local imaging and medical data for risk modeling may create more appropriate models for our region and models that are more universally applicable throughout the world. Currently, data from residents in Hawaii are not used in any known cancer risk models. For breast cancer, for example, there are only 7000 native Hawaiian or Pacific Islanders identified in the Breast Cancer Surveillance Consortium's database of over 11 million women in the United States - none of these women live in Hawaii. In this talk, I will describe the efforts going on in the UH Cancer Center to create Hawaii-specific AI and deep learning training datasets for cancer detection and risk modeling. I will also describe the deep learning methods, architectures, and questions being asked in our exploration of cancer risk using deep learning in Hawaii. I will provide several current student projects as examples in skin and breast cancer, and obesity.

Bio

Dr. Shepherd is a Researcher/Professor and an expert in mining biomarkers from medical imaging using advanced machine learning techniques. He joined the UH Cancer Center in early 2018 after an 19 year career at University of California San Francisco. He has a broad background in medical imaging, and considered the leading expert in the theory and practice of dual-energy X-ray absorptiometry (DXA). He has three NIH studies for diagnostic breast imaging and Body Shape as it relates to human health in children and adults. He directs the UHCC Artificial Intelligence Precision Health Institute, the Hawaii Pacific Island Mammography Registry, the Body Composition Exercise Physiology and Energy Metabolism CORE lab, and his own research group. He takes on translational-science graduate students interested in applying their field of study to cancer research using advanced data science methods. He currently has graduate students with home departments in Kinesiology, Electrical Engineering, Computer Science, and Nutrition Science. He has over 200 peer-reviewed publications that have been referenced over 10,000 times.