Enacting a statewide crisis response to the progression of COVID-19 requires a multiscale data collection, aggregation, and analysis strategy to circumvent variability associated with hyper-local effects of individual hospitals on one end and excessively coarse models based on national averages on the other. Consider the relationship between local health care systems in normal times. Beyond health care privacy laws and corporate rules, different hospital systems are friendly competitors, not used to sharing data about real-time functioning. In contrast, the extraordinary nature of the COVID-19 crisis required state wide coordination among Delaware hospitals to rapidly develop “shotgun” data agreements so that models can be fit with real-time data, balancing need and required effort. Integrating these data into models generates projections that inform constantly changing risk theory analyses, including the need for additional hospital facilities and medical supplies. With this proposal, we seek to develop a framework for data collection, aggregation, integration, and cooperation based on our Delaware case-study that can be used by hospital systems throughout the US to plan resources as the COVID-19 pandemic evolves as well as for future pandemics. The highly interdisciplinary project team consists of experts in Public Policy, Epidemiology, Physics, Computer Science and Industrial Engineering and has strong integration with local health systems in Delaware.
There are two prongs to this project. The first is to study homeless data for the State of Delaware; the second is to model demand for beds in the statewide hospital capacity in response to COVID-19 data from Delaware. Prof. RJ Braun, director of the MS in data science at UD, will be the overall mentor for the projects, and will coordinate the efforts led by Profs. Metraux and Dobler.