Data-Driven Patient Risk Stratification: Leveraging Data Across Time and Space to Enhance Local Care
Add to Google Calendar
The proliferation of electronic medical records holds out the promise of using machine learning and data mining to build models that will help healthcare providers improve patient outcomes. However, building useful models from these datasets presents many technical problems. The task is made challenging by the large number of factors, both intrinsic and extrinsic, influencing a patient's risk of an adverse outcome and the inherent evolution of that risk over time.
In this talk, I will describe the development and validation of a risk stratification model for predicting healthcare-associated infections (HAIs), one of the top-ten contributors to death in the US. I will show how by adapting techniques from time-series classification, transfer learning and multi-task learning one can learn a more accurate model for predicting infections with Clostridium difficile (one of the most common HAIs).
Applied to a held-out validation set of 25,000 patient admissions, our model achieved an area under the receiver operating characteristic curve of 0.81 (95%CI 0.78-0.84). On average, we can identify high-risk patients five days in advance of a positive test result. We are now working with clinicians at the hospital to identify ways in which that information can be used to reduce the incidence of HAIs. To this end, I will conclude the talk by describing our current research efforts and directions that promise to further increase the utility of such data-driven models.
Jenna Wiens is an Assistant Professor in EECS at the University of Michigan. In the fall of 2014, she joined the CSE division after completing her PhD at MIT.
Professor Wiens' primary research interests lie at the intersection of machine learning and medicine. She especially enjoys solving the technical challenges that arise when considering the practical application of machine learning in clinical settings. Currently, she is focused on developing accurate patient risk stratification approaches that leverage data across time and space, with the ultimate goal of reducing the rate of healthcare-associated infections among patients admitted to hospitals in the US.
In addition to this work, Professor Wiens had the privilege of working with a unique dataset from the NBA. Recently, she has had a lot of fun applying many of the same techniques used in the medical work to the world of sports analytics. In general, Professor Wiens enjoys tackling the challenges that develop when working with large complex datasets.