I lead the development, implementation, and evaluation of machine learning models for a variety of clinical use cases and build tooling to streamline and ensure the quality of machine learning solutions implemented into the clinical setting.

I am working with clinical faculty to integrate statistical/machine learning models into the health care setting, with use cases involving heart failure, sepsis, population health, and palliative care. These methods include novel modeling methodologies specifically tailored to a specific use-case while respecting constraints around data privacy, run-time, and model complexity. I also explore modern methods, including deep learning for medical image classification and natural language processing.I also develop software tools for clinicians and researchers to use to visualize, manipulate, and better understand Electronic Health Record data regarding their particular use case. Specifically, I developed an application aimed at hotspotting high utilizers within a particular clinic and its use has led to more efficient provisioning of limited resources to patients that benefit the most.In addition, I direct an initiative for the development of mobile applications (iOS, Android) for healthcare-related use cases as part of an effort at Duke University to leverage undergraduate computer science talent in order to tackle existing health system issues.

As an intern at the Duke Institute for Health Innovation, I developed an R Shiny app in order to identify inefficient use of health system services by leveraging the electronic health record data at Duke Health. In doing so, I was given the opportunity to work with health care providers from various clinics in order to identify high-cost patients and identify which of them would be best suited for a variety of interventions.

I am currently teaching a course that I designed that deals with the intersection of data science, machine learning, and healthcare. The course is listed as BME 590: Data Science and Health.http://dsahduke.github.io