How can we learn from the past to predict the next disease outbreak?
Machine learning is a type of artificial intelligence where computers learn from data and make predictions. Machine learning models have powerful potential for disease forecasting because they recognize patterns by analyzing past disease data, and they can use these patterns to predict when and where an outbreak might occur. Â
The CIDMATH Machine Learning team, led by Dr. Max Lau, is developing and applying advanced machine learning models to improve public health. They systematically evaluate different types of machine learning models to determine the most reliable methods for different disease patterns. This research moves beyond simple predictions to create trustworthy, interpretable systems that help officials get ahead of infectious disease outbreaks.Â
Projects
Comparative AnalysisÂ
The team has used endemic measles dynamics to explore machine learning approaches compared to and integrated with classical mechanistic models. Comparison of a neural network and classical mechanistic model demonstrates that the neural network model overall outperforms the mechanistic model across forecasting windows.  Â
Additionally, incorporating spatial features into the neural network model revealed the hierarchical spatial structure of measles spread, with major cities driving regional outbreaks.Â
Latest Works
Article: Deep neural networks for endemic measles dynamics: Comparative analysis and integration with mechanistic models. PLOS Computational Biology, 2024
Wyatt Madden, Wei Jin, Benjamin Lopman, Andreas Zufle, C. Benjamin Dalziel, Jessica Metcalf, Bryan Grenfell, Max Lau
TEAM
Max Lau, PhD
Assistant Professor, Department of Biostatistics and Bioinformatics, Emory University Rollins School of Public Health, Co-Investigator, CIDMATH
Contact
Publication Link
Emory Profile
Jiaxi Geng, MPH
PhD Student, Department of Biostatistics and Bioinformatics, Emory University Rollins School of Public Health