Prediction of influenza-like illness based on the improved artificial tree algorithm and artificial neural network

Area of Science:

• Public Health
• Computational Epidemiology
• Artificial Intelligence

Background:

• Influenza is a significant public health threat requiring accurate forecasting.
• Real-time prediction of influenza outbreaks is crucial for timely interventions.
• Existing methods may lack the speed or accuracy for real-time public health needs.

Purpose of the Study:

• To develop a novel method for nearly real-time prediction of regional influenza-like illness (ILI) percentages in the United States.
• To leverage social media (Twitter) and traditional surveillance data (CDC ILI) for improved forecasting accuracy.
• To evaluate the efficiency of an artificial neural network optimized by an improved artificial tree algorithm for this prediction task.

Main Methods:

• Utilized a combined dataset including Twitter data and United States Centers for Disease Control (CDC) influenza-like illness (ILI) data.
• Employed an artificial neural network (ANN) model for predictive analysis.
• Optimized the ANN using an improved artificial tree (IAT) algorithm to enhance prediction performance.

Main Results:

• The proposed method demonstrated efficient and accurate nearly real-time prediction of regional ILI percentages.
• Integration of Twitter data alongside CDC ILI data improved forecasting capabilities.
• The optimized ANN model proved effective in capturing influenza trends.

Conclusions:

• The developed approach offers a viable and efficient solution for real-time influenza outbreak prediction.
• This method can aid public health officials in making informed decisions during influenza seasons.
• Combining diverse data sources with advanced AI techniques enhances epidemiological surveillance.