A team of researchers at the Indian Institute of Engineering Science and Technology, Howrah, and the Indian Statistical Institute, Kolkata, has developed a mathematical model that uses local weather patterns to predict the spread of dengue fever up to two years in advance. By analysing a decade of climate and health data from Brasília, Brazil, the researchers have successfully linked the life cycle of the Aedes aegypti mosquito directly to fluctuations in temperature, rainfall, and humidity. This new approach allows health officials to detect an epidemic weeks before the first human cases peak, potentially saving thousands of lives through more timely mosquito control.

To build their model, the team relied on mechanistic modelling. Instead of examining only how many people became ill in the past, the researchers built a six-part virtual world that simulates the mosquito’s entire life cycle. They tracked the aquatic phase, in which eggs develop into larvae in puddles of rainwater, and the adult phase, in which female mosquitoes bite humans to transmit the virus. By using mathematical equations to represent these stages, the team could see exactly how a spike in temperature speeds up mosquito development or how a specific amount of rainfall creates the perfect breeding ground.

To improve the model’s predictive performance, the team integrated machine learning techniques, including Neural Networks and Long Short-Term Memory (LSTM) models. These AI tools were trained on data from 2014 to 2021 and then tested against the 2022 and 2023 outbreaks. The results showed that the model’s Basic Reproduction Number, a mathematical measure of how many new people one infected person will infect, closely matched the real-world rises and falls in dengue cases. The researchers found that while temperature is a major driver of mosquito biting rates, moderate rainfall provides them with the optimal conditions for breeding. Excessive rainfall tends to flush away the larvae, while insufficient rainfall dries up their habitats.

The study concludes that the most effective way for society to stop dengue is a vector-first approach. The mathematical simulations showed that treating sick humans alone is insufficient to stop an epidemic because the reproduction number remains high as long as mosquitoes continue to breed in the streets. To effectively control an outbreak, the data suggest that intensive mosquito control, such as removing stagnant water, must be combined with medical treatment. By providing a two-year look ahead, this research gives cities a vital window to clean up breeding sites and prepare hospitals before the mosquitoes even hatch.

However, the researchers noted that their model primarily focuses on environmental drivers and does not yet account for socioeconomic factors, such as sudden changes in human mobility or variations in how different neighbourhoods report medical cases. For accurate prediction, these factors must be accounted for.

Nonetheless, many older models often focused solely on the biology of the mosquito’s lifespan or its environment. By making these factors climate-sensitive, this study provides a clever way of globally tracking the disease without tracking the mosquitoes at any particular location. In a world where climate change is expanding the reach of tropical diseases, this fusion of biology, math, and AI offers a vital shield. It turns unpredictable weather into a predictable map for public health, ensuring that resources are sent exactly where they are needed most.