RAPID: CCF: Optimizing Resource Allocation to Combat Pandemics

Computer and Information Science and Engineering - Network mobility models are important in the analysis of the COVID-19 pandemic, and are especially useful for optimizing allocation of resources to combat the spread of infections. The recent pandemic highlights this, as well as the need for methods to determine timely decisions for strategic interventions that reduce the impact of the pandemic on populations. Use of network traffic models account for flow of the disease via carriers from the initial source of the pandemic and between centers of infections, and addresses the long-distance spread of the disease. Non-medical solutions that immediately attempt to reduce the spread of pandemic include intra-city restrictions and inter-city strategies that involve suppression of population transfer. Critical actions include decisions on the level of suppression, the routes over which suppression has to be applied, and the time at which it has to be applied. Reducing the mitigation or suppression must critically account for the re-occurrence of the disease. The level of suppression has economic consequences with immediate and potential long term impacts on employment and economic growth, and can run counter to maintaining essential services such as food distribution, medical facilities, and first responders. This project will develop novel techniques to analyze pandemic models and design new optimization algorithms that provide decision strategies, accounting for costs, including the economic costs of suppression. This research has time urgency as there is a need for strategic analysis in the current pandemic and the project will utilize timely insights from the data available. Additionally, the insights will assist in determining decision strategies for future occurrences.

This RAPID project will identify and refine network models of pandemic spread using a hierarchical model that incorporates the traffic network between major cities and countries at the first level. The subsequent levels will utilize local traffic networks and mobility patterns in centers of large populations. The models will include (i) subdivision of populations into classes that represent the current state of the pandemic, examples being population sets that are susceptible, infected, suppressed and recovered, all parameterized by time, (ii) multiple source and destination network flow models of infection flow, and (iii) geographic models of infection spread in local population clusters. This project will apply optimization techniques and network analysis to analyze the models and design algorithms for determining decision strategies. Evolution of the population sets, as modeled by differential equations solved using numerical methods and discrete analogs, will be investigated. Methods to determine parameters that regulate the transfer rates between populations will be designed. The model will be used to define mathematical programs in order to optimize decision parameters that include the level of suppression and the time at which to relax suppression. Network flow techniques will be used to minimize the flow of infection with multiple key objectives, especially to minimize the peak levels of the spread.

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.