Predicting COVID-19 Cases From Atmospheric Parameters Using Machine Learning Approach

Area of Science:

• Epidemiology
• Environmental Science
• Computer Science

Background:

• The dynamic spread of COVID-19 necessitates advanced predictive modeling.
• Understanding environmental and historical data's role in transmission is crucial for control.

Purpose of the Study:

• To forecast COVID-19 cases using past infection data.
• To predict current COVID-19 cases by incorporating environmental factors like PM2.5, temperature, and humidity.
• To evaluate the efficacy of different machine learning classifiers for these predictions.

Main Methods:

• Utilized four machine learning classifiers: Decision Tree, K-nearest neighbor (KNN), Support Vector Machine (SVM), and Random Forest.
• Employed Root Mean Square Error (RMSE) to assess model performance.
• Analyzed historical case data and environmental parameters (PM2.5, temperature, humidity).

Main Results:

• K-nearest neighbor and Support Vector Machine algorithms demonstrated superior performance in forecasting COVID-19 cases based on historical data.
• Temperature emerged as the most significant predictor, followed by relative humidity, for current COVID-19 case numbers.
• Decision Tree models exhibited lower predictive accuracy when using particulate matter and atmospheric conditions.

Conclusions:

• Machine learning offers a viable approach for predicting virus infections, including COVID-19.
• Environmental factors, particularly temperature, play a significant role in COVID-19 transmission dynamics.
• These predictive capabilities can empower policymakers for proactive public health interventions and monitoring.