Machine learning in epidemiology: Neural networks forecasting of monkeypox cases

Area of Science:

• Epidemiology
• Computational Biology
• Public Health

Background:

• Recent monkeypox outbreaks necessitate advanced forecasting tools.
• Traditional epidemiological models may not capture complex transmission dynamics.
• Machine learning offers novel approaches to disease surveillance.

Purpose of the Study:

• To evaluate Artificial Neural Networks (ANNs), Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU) models for monkeypox outbreak forecasting.
• To compare the predictive performance of these machine learning models in Canada, Spain, the USA, and Portugal.
• To identify optimal model configurations for effective infectious disease prediction.

Main Methods:

• Utilized ANNs, LSTM, and GRU models for forecasting.
• Trained models on monkeypox case data from June 3 to December 31, 2022.
• Evaluated model performance against test data from January 1 to February 7, 2023.
• Employed the Levenberg-Marquardt algorithm for model training optimization.

Main Results:

• ANN models demonstrated high effectiveness in forecasting monkeypox outbreaks.
• Optimized models achieved favorable Root Mean Squared Error (RMSE), Mean Absolute Percentage Error (MAPE), and Coefficient of Determination (R²) values.
• Comparative analysis highlighted the superior predictive capabilities of certain ANN configurations.

Conclusions:

• Machine learning, particularly ANNs, shows significant promise for infectious disease forecasting.
• Optimized ANN models can substantially improve public health responses to outbreaks.
• The study provides valuable insights into the application of neural networks in epidemiology.