TY  - JOUR
TI  - Augmented Artificial Neural Network Model for the COVID-19 Mortality Prediction: Preliminary Analysis of Vaccination in Turkey
AB  - The spread and severity of coronavirus disease 2019 (COVID-19) have a severe impact on our lives, so that over 4.6 million lives have been lost since it has been first emerged. Although prediction of the COVID-19 mortality may be inevitably accompanied by uncertainty, it is helpful for health politicians and public health decision-makers to take proper precautions to diminish the pandemic's severity. Therefore, this study proposed a mortality prediction model for the deaths that occur on-day, lag 1 day, lag 7 day, and lag 14 day in Turkey, considering 16 variables under four categories as follows: (i) severity of the disease, (ii) vaccination policy as a preventive strategy, (iii) exposure duration in society, (iv) time series impact. The developed Augmented- Artificial Neural Network (ANN) model took advantage of Auto-Regressive Integrated Moving Average (ARIMA) and ANN models to capture the linear and nonlinear components of the mortality. The proposed model was able to predict mortality with the lowest error compared to ARIMA and ANN models. To reveal the impact of each responsible category on mortality, a set of experiments was designed. According to the experiments' results, it was observed that the impact of four categories from highest to the lowest importance on prediction performance were exposure duration in society, vaccination policy, severity of disease, and time series, respectively. According to these results, new virus-fighting policies can be developed, and the existing model can be used as a simulation tool with the new data to be obtained.
AU  - Gunay, Elcin
AU  - Kır, Sena
DO  - 10.35377/saucis.05.01.999373
PY  - 2022
JO  - Sakarya University Journal of Computer and Information Sciences  (Online)
VL  - 5
IS  - 1
SN  - 2636-8129
SP  - 22
EP  - 36
DB  - TRDizin
UR  - http://search/yayin/detay/514586
ER  -