RISK ANALYSIS AND PREDICTION FOR COVID19 DEMOGRAPHICS IN LOW RESOURCE SETTINGS USING A PYTHON DESKTOP APP AND EXCEL MODELS.

Ali Kyagulanyi; Joel Tibabwetiza Muhanguzi; Oscar Dembe; Sheba Kirabo

Governments around the world have found it hard to estimate the potential burden of the pandemic, hence countries like Uganda have not yet fully established a contingency plan based on statistical and mathematical predictions. This has led to low efforts in expanding and sustaining investment to build preparedness and health capacity in Uganda. Using Italy and South Korea as an example, South Korea confirmed its first case on 20 th January (Wikipedia 2020) and by 22 nd march South Korea has 2909 cases with only 104 deaths. On the other hand, Italy confirmed its first cases on 31 st January 2020 and has 86498 cases with 9314 deaths by 28 th march 2020 (University 2020) . This was due to under planning for the pandemic which has registered the most severe effects compared to a country like South Korea which planned resources based on mathematical and computational predictions. South Korea employed mathematical predictions like kinetic model which was used to simplify the SIR model to plan (GeneOnline 2020), for the number of patients and make key decisions on quarantines' and city lock down , For South Korea this seems have paid off as they have managed to steadily suppress the pandemic. This paper presents a risk analysis and statistical prediction Desktop application and excel model of covid19 disease to the populations in low resource settings countries like Uganda. It predicts the trends in case of an outbreak for a period of at most 60 days from onset. The algorithm in the application employs a mathematical model of an Ordinary Differential Equation representing the SIR model (susceptible, infected and recovered (or removed) which has been used by epidemiologists in the past to predict the number of individuals likely to be infected, recover or die in the event of any epidemic. The models give results for expected number of infections in a single susceptible district like Kampala. Using infection force from different countries we sampled two rates of infection 0.29 and 0.594 for moderate and worst-case scenarios and generated epidemiological curves showing the trend the disease is expected to follow. Similar models have been designed to predict the fate of the disease in countries like united states of America, United Kingdom and South Korea and the have been found to have a high predictability. The application accepts inputs of force of infection, recovery, number of people in an area, initial number of people infected and recovered from the population. Using the python programming language, calculus and condition probability we have designed the algorithm that generates epidemiological curves and parameter (SIR) tables as an output, the efficiency of the application has been tested with a similar model implemented in an excel worksheet, predictions and graphs have been made from excel.