This Faculty Early Career Development Program (CAREER) award will contribute to the advancement of the national health and welfare by designing effective data-driven methods for personalized management of chronic diseases. Increased incidence and prevalence of chronic conditions such as diabetes and obesity have led to unsustainable growth in U.S. healthcare spending and have significantly impacted well-being. Currently, chronic diseases are managed by choosing treatments based on average patient outcomes in clinical studies with relatively homogeneous patient populations. While it is well-recognized that an individual's disease trajectory is influenced by patient-specific traits, personalized treatment has been limited by difficulties in disentangling the complex relationships between treatment plans, patient characteristics, disease prognosis, and adherence to treatments. This award will support the development and validation of algorithms and models that combine data from large patient cohorts and from specific patients in order to bring management of chronic diseases into the home setting. The educational components of this award will support development of hands-on teaching modules and research mentoring opportunities, with the aim of broadening interest in operations research and its impact on problems in healthcare, particularly among underrepresented groups.
This research will develop and analyze new tools to identify causal relationships from medical data, develop and analyze new tools to personalize treatments based upon identified causal relationships while considering dynamic and stochastic variations, and validate effectiveness of these tools through simulation studies based on medical datasets of diabetes and weight loss patients. The approach will be to develop new statistical and optimization methods that identify causal relationships between treatment plans, patient characteristics, disease prognosis, and adherence to treatments; and then use these estimated causal relationships to personalize treatments. This research goes beyond existing approaches that identify only correlations in data and which generally lead to sub-optimal treatment plans. Anticipated outcomes of this research include new statistical and optimization models, new theoretical analyses, and a clinically-relevant understanding of the strengths and limitations of the developed models. Success in this project has the potential to improve health outcomes and reduce healthcare costs through increased personalization of treatment and automation of healthcare delivery.
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
