Forecasting COVID-19 Cases Using Alpha-Sutte Indicator: A Comparison with Autoregressive Integrated Moving Average (ARIMA) Method.

COVID-19 is a pandemic which has spread to more than 200 countries. Its high transmission rate makes it difficult to control. To date, no specific treatment has been found as a cure for the disease. Therefore, prediction of COVID-19 cases provides a useful insight to mitigate the disease. This study aims to model and predict COVID-19 cases. Eight countries: Italy, New Zealand, the USA, Brazil, India, Pakistan, Spain, and South Africa which are in different phases of COVID-19 distribution as well as in different socioeconomic and geographical characteristics were selected as test cases. The Alpha-Sutte Indicator approach was utilized as the modelling strategy. The capability of the approach in modelling COVID-19 cases over the ARIMA method was tested in the study. Data consist of accumulated COVID-19 cases present in the selected countries from the first day of the presence of cases to September 26, 2020. Ten percent of the data were used to validate the modelling approach. The analysis disclosed that the Alpha-Sutte modelling approach is appropriate in modelling cumulative COVID-19 cases over ARIMA by reporting 0.11%, 0.33%, 0.08%, 0.72%, 0.12%, 0.03%, 1.28%, and 0.08% of the mean absolute percentage error (MAPE) for the USA, Brazil, Italy, India, New Zealand, Pakistan, Spain, and South Africa, respectively. Differences between forecasted and real cases of COVID-19 in the validation set were tested using the paired t-test. The differences were not statistically significant, revealing the effectiveness of the modelling approach. Thus, predictions were generated using the Alpha-Sutte approach for each country. Therefore, the Alpha-Sutte method can be recommended for short-term forecasting of cumulative COVID-19 incidences. The authorities in the health care sector and other administrators may use the predictions to control and manage the COVID-19 cases.

Phenomenological Modelling of COVID-19 Epidemics in Sri Lanka, Italy, the United States, and Hebei Province of China.

The COVID-19 pandemic has resulted in increasing number of infections and deaths every day. Lack of specialized treatments for the disease demands preventive measures based on statistical/mathematical models. The analysis of epidemiological curve fitting, on number of daily infections across affected countries, provides useful insights on the characteristics of the epidemic. A variety of phenomenological models are available to capture the dynamics of disease spread and growth. The number of daily new infections and cumulative number of infections in COVID-19 over four selected countries, namely, Sri Lanka, Italy, the United States, and Hebei province of China, from the first day of appearance of cases to 2nd July 2020 were used in the study. Gompertz, logistic, Weibull, and exponential growth curves were fitted on the cumulative number of infections across countries. AIC, BIC, RMSE, and R 2 were used to determine the best fitting curve for each country. Results revealed that the most appropriate growth curves for Sri Lanka, Italy, the United States, and China (Hebei) are the logistic, Gompertz, Weibull, and Gompertz curves, respectively. Country-wise, overall growth rate, final epidemic size, and short-term forecasts were evaluated using the selected model. Daily log incidences in each country were regressed before and after the identified peak time of the respective outbreak of epidemic. Hence, doubling time/halving time together with daily growth rates and predictions was estimated. Findings and relevant interpretations demonstrate that the outbreak seems to be extinct in Hebei, China, whereas further transmissions are possible in the United States. In Italy and Sri Lanka, current outbreaks transmit in a decreasing rate.