Forecasting adversities of COVID-19 waves in India using intelligent computing.

Area of Science:

• Epidemiology
• Data Science
• Machine Learning

Background:

• The second wave of COVID-19 in India caused widespread devastation, necessitating accurate forecasting models.
• Millions were affected, with ongoing recovery challenges, highlighting the need for predictive tools to manage the epidemic.

Purpose of the Study:

• To develop and evaluate robust forecasting models for cumulative COVID-19 infections in India.
• To assess the accuracy of various machine learning methods for short-term epidemic prediction.
• To identify the most accurate model for aiding public health officials in hazard mitigation.

Main Methods:

• Employed seven forecasting methods: Auto-ARIMA, Auto-ETS, Auto-MLP, Auto-ELM, AM, MLP, and a proposed Extreme Learning Machine (ELM) model.
• Conducted 90-day advanced forecasting for cumulative COVID-19 cases up to July 24, 2021, using 15-day intervals.
• Fine-tuned model hyperparameters to optimize prediction performance and extracted five features to understand dataset characteristics.

Main Results:

• The proposed Extreme Learning Machine (ELM) model achieved a Mean Absolute Percentage Error (MAPE) of 5.01%.
• The ELM model demonstrated superior accuracy compared to the other six evaluated forecasting methods.
• Feature extraction provided insights into the dataset, supporting further model investigation with updated data.

Conclusions:

• The proposed ELM model offers a reliable and accurate approach for forecasting cumulative COVID-19 cases in India.
• This predictive capability can assist health officials in managing and mitigating the impact of infectious disease outbreaks.
• Further research with updated datasets is encouraged to validate and refine the proposed model's performance.