Predicting Hospital Beds Utilization For COVID-19 In Telangana, India

Darapaneni, N.; Mahita, G. M.; Paduri, A. R.; Talupuri, S. K.; Konanki, V.; Galande, S.; Tondapu, C. H.

Our study aims to predict the requirement of isolation beds, oxygen beds, and Intensive Care Unit (ICU) beds in both government and private hospitals for COVID-19 patients in Telangana, India. Using epidemic modeling and time series analysis techniques, we can predict the future daily COVID-19 cases and active cases from historical data of confirmed COVID-19 cases. From the historical data analysis, we can also establish the pattern of hospitalized cases from active cases and also the pattern of various categories of bed occupancy numbers. The pattern is represented by minimum, maximum and average percentages. We have observed the second wave of COVID-19 in many European countries and is spreading more rapidly than the first wave. In India, we must be prepared to address the second wave as it might be more rampant since the lockdown is eased in most of the cities. Since traveling and gatherings are getting back to normal, significantly larger population might be susceptible and would need immediate hospitalization. We presented a process/methodology to estimate bed occupancy numbers from the prediction of COVID-19 positive and active cases using modified Susceptible-Infected-Recovered (SIR) model for the contagion and FB-Prophet model for time series analysis. We used official data from the Government of Telangana bulletins for the COVID-19 pandemic up to December 31st, 2020 for our process/methodology. SIR modeling is more intuitive and explainable but requires a lot of trial and error and assumptions. The FB-Prophet prediction process is simple and accuracy is also better compared to SIR modeling.