Predicting Cardiovascular Outcomes with CRADLE
Dr. Joyce Ho is an assistant professor in the Computer Science Department at Emory University. Her NIH funded research focuses on leveraging modern machine learning algorithms to predict cardiovascular complications in patients with diabetes. Early identification of patients at high-risk of developing cardiovascular complications is crucial for providing effective interventions. To build accurate and robustmachine learning models to identify high-risk patients, a large patient population is needed.
By taking advantage of the Emory Clinical Research Analytics Data Lake Environment (CRADLE), Dr. Ho and her collaborators have been able to analyze existing scoring systems and machine learning tools on close to 48,000 patients with type 2 diabetes seen at Emory Healthcare System between 2013 and 2017. Her experience with the AWS at Emory team has been superb. "I have the flexibility to personally manage the AWS resources, making it seamless for me to launch new instances and run computationally intensive jobs. The team has also been very helpful with the transition from the regular AWS platform."
The biggest game changer though in her opinion is CRADLE. "I can query over 8 million Emory patients using Athena to perform exploratory analyses. Using CRADLE, I have been able to quickly obtain study population statistics, determine what data is captured, and even get some preliminary results for several grant proposals. The Data Solutions Team has also been extremely supportive. When NYU and Emory collaborators wanted to study the prevalence of diabetes in South Asians, we were able to accomplish this in CRADLE."
AWS at Emory is primed to support big data analytics and machine learning projects for researchers like Dr. Ho. With new tools and services regularly being added to the platform as well as a dedicated team to support her efforts, the sky truly is the limit for Dr. Ho's work.