Isfahan and Covid-19: Deep spatiotemporal representation

Area of Science:

• Epidemiology
• Public Health
• Data Science

Background:

• COVID-19 significantly impacted global health, with Iran being an early affected country.
• Isfahan province experienced a notable COVID-19 epidemic, necessitating accurate forecasting for policymaking.

Purpose of the Study:

• To forecast the COVID-19 epidemic size, peak value, and peak time in Isfahan.
• To evaluate the impact of Social Determinants of Health (SDH) on epidemic prediction.
• To develop a deep learning model for accurate spatiotemporal forecasting.

Main Methods:

• Utilized deep learning, specifically a tailored Long Short-Term Memory (LSTM) network.
• Integrated spatiotemporal COVID-19 data with Social Determinant of Health (SDH) factors.
• Incorporated mutual effects of confirmed, death, and recovered cases in the prediction model.

Main Results:

• The tailored LSTM model demonstrated high accuracy in both short- and long-term COVID-19 forecasting.
• The inclusion of SDH factors significantly improved pandemic prediction capabilities.
• The model highlighted the importance of considering all case classes (confirmed, death, recovered) for accurate predictions.

Conclusions:

• The proposed deep learning model, incorporating SDH, is a reliable tool for health decision-makers in Isfahan.
• SDH factors play a crucial role in enhancing the accuracy of epidemic forecasting.
• The model's long-term prediction ability is a key strength for strategic health planning.