Data Science Fellowship Program

The Duke AI Health Data Science Fellowship Program is a 2-year training program in data science with direct application for healthcare. Designed for early-career data scientists with strong backgrounds in quantitative discipline, the program is part of a multidisciplinary, campus-spanning initiative that applies machine learning and quantitative sciences to rich sources of healthcare and administrative data, using the insights gained to improve healthcare delivery, quality of care, and the health of individuals and communities.

Data Science Fellows are integrated into multidisciplinary teams, where they are mentored by and work under the direction of the team’s quantitative lead, who offers advanced expertise in data science and machine learning. Fellows also receive guidance and mentoring from the team’s clinical lead(s), who provide insight into the clinical and operational contexts in which projects are embedded, and in some cases receive mentoring from a staff statistician. Data science teams also include a project manager and may include an informaticist.

Each team focuses on a well-defined research question or problem related to healthcare delivery, quality of care, or clinical operations. Working in this intense but supportive environment, trainees acquire essential skills in healthcare analytics and modern machine learning methods, while also benefiting from in-depth exposure to real-world clinical problem-solving. The Data Science Fellow works as a fully integrated member of these multidisciplinary teams, which are assembled jointly with input from AI Health, the Department of Biostatistics and Bioinformatics, Duke University Health System, and Duke School of Medicine clinical leadership.

AI Health Fellows typically work on 2 to 3 projects each year and are encouraged to collaborate across complementary projects as opportunity and topic focus align.

Meet the AI Health Fellows


Duke AI Health to Collaborate with Microsoft’s Azure platform

In collaboration with Microsoft, Duke AI Health is working to establish an analytic environment based on Microsoft’s Azure cloud computing platform to support the ongoing development of predictive modeling applications at Duke University. An initial pilot effort is focused on configuring the environment to support common data science workflows using a publically available dataset from the 2012 PhysioNet  Computing in Cardiology Challenge (‘Predicting Mortality in ICU Patients’). This dataset was selected because many of its features are consistent with the demands of activities related to health data science at Duke. The Azure environment provides an
opportunity for research fellows and students to acquaint themselves with the Azure machine learning service offerings in a collaborative setting.

In the second phase of this pilot, we plan to expand the workflows into a format that complies with standards that govern the use of protected health information (PHI), thus enabling current research to be conducted within the Azure infrastructure. These efforts represent a significant step toward cloud-based machine learning at Duke and provide an excellent training opportunity for AI Health Fellows and other learners. We are grateful to the Microsoft team for their support of the AI Health Fellowship program and this initiative