The outbreak of the coronavirus disease-19 (COVID-19) pandemic has created much speculation on the behavior of the disease. Some of the questions that have been asked can be addressed by computational modeling based on the use of high-performance computing (HPC) and machine learning techniques.
The Reference Model previously used such techniques to model diabetes. The Reference Model is now used to answer a few questions on COVID-19, while changing the traditional susceptible-infected-recovered (SIR) model approach. This adaptation allows us to answer questions such as the probability of transmission per encounter, disease duration, and mortality rate. The Reference Model uses data on US infection and mortality from 52 states and territories combining multiple assumptions of human interactions to compute the best fitting parameters that explain the disease behavior for given assumptions and accumulated data from April 2020 to June 2020.
This is a preliminary report aimed at demonstrating the possible use of computational models based on computing power to aid comprehension of disease characteristics. This infrastructure can accumulate models and assumptions from multiple contributors.