Unraveling the drivers of leptospirosis risk in Thailand using machine learning

What is it about?

The research team built an artificial intelligence model using a technique called XGBoost to analyze 16 years of disease surveillance data alongside information about rainfall, temperature, rice farming activities, household size, and economic indicators across Thailand's provinces. They trained the model on data from 2007-2017 to distinguish between provinces with high versus low disease risk, then tested its accuracy on data from 2018-2022. To understand which factors were most important for predictions, they used a method called SHAP analysis that reveals how each factor influences the model's decisions.

Featured Image

Photo by Conny Schneider on Unsplash

Photo by Conny Schneider on Unsplash

Why is it important?

Leptospirosis is a serious bacterial disease that poses a major public health threat in Thailand, with over 51,000 cases reported between 2007 and 2022. The bacteria thrive in contaminated water and soil, particularly in rice farming areas, and human infection occurs through contact with these contaminated environments. Since the disease results from complex interactions between weather patterns, farming practices, and socioeconomic conditions, researchers aimed to develop a machine learning model that could predict which provinces in Thailand are at high risk for leptospirosis outbreaks by analyzing these multiple factors together.

Perspectives