Modeling the Impact of Care Interventions on Patients with Complex Medical and Social Needs

University Of Massachusetts Amherst, Amherst MA

Investigators

Abstract

This award will contribute to the advancement of national health and welfare by studying the impact of care interventions for patients who experience significant medical, behavioral health, and social challenges. Although representing 1-5 percent of the population, such patients often experience extreme patterns of healthcare utilization and can account for 25-50 percent of national healthcare costs. Holistic, person-centered care interventions, often led by a multidisciplinary team consisting of nurses, community health workers and social workers, have emerged as a strategy to engage with and help improve the health and wellbeing of such patients. This project studies several open questions, including longitudinal patterns of healthcare use among such patients, the role of specific combinations of medical and social factors, and the impact of the time spent by the care team on patient outcomes. The project comprises a collaboration between the University of Massachusetts, Amherst, and the Camden Coalition of Healthcare Providers, an organization with significant experience in care interventions for patients with complex medical and social needs. The accompanying dissemination plan will introduce results to practitioners in the domain through conferences, undergraduate and graduate students through research projects and courses, and the general audience through non-fiction essays. The primary contribution of this research is the adaptation of stochastic methodologies to patient-level longitudinal data that spans multiple parts of health and social services systems, including such data from Camden Coalition’s multi-year randomized controlled trial (RCT). The project has two aims. In Aim 1, the goal is to create stochastic models of event progression to model the sequence of timing of key events (e.g., intervention enrollment, post-discharge primary care visit, hospital readmissions, social milestones) on the patient’s timeline. The sequence and timing models will be used in conjunction with search algorithms that identify subgroups in the covariate space (defined by patient demographics, medical conditions and social needs) where intervention and control groups in the RCT differ significantly. In Aim 2, the project will create a multi-period stochastic decision process that integrates staffing and prioritization decisions with a risk estimation function to evaluate how patient outcomes such as hospital readmissions are impacted. Thus, the project aims to bring together a unique mix of data-driven stochastic, optimization and statistical learning methodologies informed by longitudinal data to a vital yet under-explored domain of healthcare. 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.