Feature selection based multivariate time series forecasting: An application to antibiotic resistance outbreaks...

Area of Science:

• Infectious Disease Epidemiology
• Computational Biology
• Health Informatics

Background:

• Antimicrobial resistance is a critical global health challenge requiring proactive prevention strategies.
• Predicting infectious disease outbreaks is essential for effective public health interventions and resource allocation.

Purpose of the Study:

• To develop and evaluate an Artificial Intelligence (AI) based approach for forecasting Methicillin-resistant Staphylococcus aureus (MRSA) outbreaks.
• To apply Time Series Forecasting techniques combined with feature selection for predicting MRSA incidence.

Main Methods:

• Utilized multivariate time series data including Methicillin-sensitive and MRSA infections, influenza incidence, and antimicrobial therapy days (Levofloxacin, Oseltamivir) from 2009-2018.
• Employed wrapper feature selection methods driven by multi-objective evolutionary algorithms (MOEA) and state-of-the-art regression algorithms.
• Implemented a novel multi-criteria decision-making process integrating performance metrics (RMSE, MAE) and prediction slopes for model selection.

Main Results:

• The proposed AI methodology successfully forecasted MRSA outbreaks with low Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) values.
• The best model achieved RMSE values of (0.1349, 0.1304, 0.1325) and MAE values of (0.1003, 0.096, 0.0987) for 1, 2, and 3 steps-ahead predictions, respectively.
• Demonstrated the efficacy of feature selection and multi-criteria decision-making in developing accurate antibiotic resistance forecasting models.

Conclusions:

• This study presents the first feature selection-based multivariate time series methodology for antibiotic resistance forecasting.
• The developed AI approach offers a promising tool for the early prediction of MRSA outbreaks, contributing to antimicrobial resistance containment efforts.
• The findings support the integration of advanced AI techniques into public health surveillance for infectious diseases.