The spread and mortality of dengue disease is intricately related to the changes in temperature, rainfall and relative humidity which can increase mortality from the mosquito borne disease.
But the possibility of dengue outbreaks can also be predicted more than two months in advance using a combination of artificial intelligence, machine learning and weather models, according to a new study published in the journal Scientific Reports on January 21, 2025.
The study led by Sophia Yacob and Roxy Mathew Koll from the Indian Institute of Tropical Meteorology (IITM), Pune, analysed data on temperature, rainfall, relative humidity and mortality from dengue from 2004-2015 for Pune city using artificial intelligence and machine learning models.
The researchers found that “warm temperatures above 27°C and humidity levels between 60 per cent and 78 per cent elevate the risk of dengue”. Additionally, they found that moderate rainfall (adding up to 150 millimetres in a week) spread throughout the southwest monsoon season (June to September) increased the mortality from dengue.
On the contrary, "heavy rains above 150 mm in a week reduce the dengue prevalence by flushing out the mosquito eggs and larvae”, according to a press release issued by the researchers.
Specifically for Pune city, the ideal temperature range for increase in dengue cases is 27°C to 35°C. The heightened temperatures aid in many factors crucial for transmission of the disease such as the lifespan of mosquitoes, their production of eggs, frequency with which they lay eggs and the time between feeding and laying eggs.
The heat also positively aids the virus’s development inside the mosquito and the time it takes for the disease symptoms to appear in humans after infection.
The scientists also highlighted that this temperature window is specific for Pune and for other places similar analysis needs to be carried out using long-term and local health and meteorological data.
The rainfall during the southwest monsoon shows high variability within the season with some periods of more rainfall known as active (wet) phases and some periods of relatively less rainfall known as break phases (dry). The number of these active and break phases determines the variability of most of India’s primary rainfall season.
The study’s analysis showed that for Pune city, low monsoon variability led to more number of dengue cases and deaths while high monsoon variability led to lower number of dengue cases and deaths. This means that “years with high dengue mortality in Pune are associated with moderate rainfall distributed over time”, according to the press release.
“In summary, it is not the cumulative amount of rainfall, but rather the pattern of rainfall, that plays a crucial role in influencing dengue transmission in Pune,” the researchers added.
The number of deaths due to dengue in Pune could further increase by 13 per cent in the near term (2020-2040) and 25-40 per cent in the mid term (2040-2060) under moderate to high greenhouse gas (GHG) emission scenario and by up to 112 per cent in the long term (2081-2100) under high GHG emission scenario.
This would be due to rising temperatures and variability of rainfall during the southwest monsoon season and without timely interventions, according to the study authors.
The 2020-2040 scenario accounts for a global average warming of 1.5°C since the pre industrial period and the 2040-2060 scenario accounts for warming of 2°C.
The scenarios come with the caveat that they are based on the future climatic conditions but not on the future socio economic factors which would also influence the transmission of dengue.
Dengue future projections for Pune
The main driver for the future increase in dengue deaths would be warmer temperatures, according to the study. Both temperature and humidity over India is set to rise similar to other places in the world and patterns of rainfall during the monsoon season are going to become even more erratic.
The occurrence of dengue can be predicted based on the India Meteorological Department’s (IMD) extended range forecasts which give probable periods of active and break phases of the monsoon 10-30 days in advance for the entire country.
This data combined with the fact that it requires some time for variations in temperature, humidity and rainfall to begin affecting dengue cases and fatalities in particular areas, which can be utilised to develop an early warning system for dengue in the nation.
“The team developed a model based on artificial intelligence / machine-learning for dengue predictions, offering more than two months of lead time for dengue outbreak preparedness,” said the press release.
The artificial intelligence model basically takes all the health, weather and climate data as input and builds the most likely scenarios for the future as forecasts of dengue occurrence and mortality.
“This can give adequate time for the local administration and health department to enhance preparedness and response strategies, potentially reducing dengue cases and deaths,” the researchers added.
For the early warning model to work it needs to be localised and for this, location-specific health data is extremely important along with the meteorological data which is not always readily available. This is in addition to the fact that dengue cases in India are grossly underreported.
“One study indicates that dengue is significantly underreported in India, revealing that the actual number of cases is 282 times higher than the reported figures,” said the study authors.
“We were able to conduct this study and prepare an early warning system using health data shared by Pune’s health department,” said Koll. “We approached Kerala and other states where dengue cases are high, but health departments there did not cooperate,” he added.
“This collaboration highlights the importance of bringing together expertise from diverse fields to address complex climate-health challenges. It is a perfect example of how scientists, the health department, and the government can work together to improve our health warning system,” said Sujata Saunik, chief secretary of the government of Maharashtra.
“We have meteorological data readily available from the IMD. If health data is shared, we can prepare customised early warning systems for climate sensitive diseases like dengue, malaria, and chikungunya for each city or district in India. Cooperation from health departments is key to saving lives,” Koll emphasised.
“States like Kerala, Maharashtra, West Bengal, Karnataka, Tamil Nadu, Gujarat, Punjab, Haryana, Andhra Pradesh, Telangana, Rajasthan and Uttar Pradesh, which bear a significant dengue burden, can greatly benefit from an advanced early warning system like this to enhance preparedness and reduce the disease's impact,” concluded the authors.