Australian dengue fever outbreaks are often traced to travellers returning from Asian tourist destinations, but an early warning system being trialled in Vietnam could substantially reduce that risk.

Australian and US researchers are working on a system using artificial intelligence to predict dengue fever danger zones in Vietnamese communities so preventive measures can be taken to curb outbreaks.

The mosquito-borne virus has no specific vaccine and there are an estimated 300 to 400 million infections annually in tropical countries, with up to 30,000 deaths.

The virus is climate-sensitive so the predictive model takes into account rainfall, temperature and humidity as well as population density and water storage practices in specific communities.

It was described by Southern Cross University academic Vinh Bui as a "highly sophisticated weather forecast" but for dengue, to identify locations with the highest outbreak risk.

The early warning system is being rolled out across districts in Vietnam's Mekong Delta in early 2026.

A colour-coded system will alert community health workers to the need to prepare for outbreaks by having residents empty or cover water storage containers where mosquitoes could lay their eggs.

Other safeguards to prevent dengue fever infection are to cover up and use insect repellent to avoid mosquito bites.

Dr Bui told a media briefing on Thursday the aim was to move from reactive to active responses to reduce the severity of outbreaks and save lives.

Northern Queensland recorded cases of dengue fever annually and funding is being sought to adapt the predictive model for that region, he said.

The model is hoped to reduce infections by 25 per cent when up and running, said academic Dung Phung from the University of Queensland's health faculty.

Three-month forecasts are the hope of Yale School of Public Health epidemiologist Dan Weinberger, with a test case that can be employed elsewhere in the region and become a model for others to follow.