In 2017, the war-torn country of Yemen suffered from the largest and fastest proliferating cholera epidemic in modern history, with as many as 50,000 new cases per week. UNICEF estimated that one child died every 10 minutes. The World Health Organization (WHO) said that there were 1.1 million cases and at least 2,300 deaths by the year’s end.

As reported by BBC News, today, there are around 2,500 cases per week. This clearly dramatic change is largely thanks to a new computer system that can accurately predict where outbreaks will occur, allowing medical professionals on the ground to nix them before they get any worse.

The key? Keeping an eye on the weather.

The situation in Yemen has only worsened since the outbreak of hostilities in 2015. Cholera, a waterborne disease, is spreading easier thanks to the violent conflict that’s savaging the country: an already poor sewage and sanitation infrastructure is being destroyed by bombs.

A two-dose, relatively affordable vaccine exists, and various rehydration and antibiotic treatments are available, but getting it to those in need isn’t always easy. Dealing with cholera in countries with poor sanitation alone is troublesome enough, but aid workers in Yemen are constantly risking their lives to do just that.

The Vibrio cholerae bacteria spreads incredibly quickly, though, which makes stemming its emergence and spread incredibly difficult. Having a way to predict where the outbreaks will occur would greatly assist them in their endeavors.

Importantly, cholera is aided by the presence of rainfall. When it’s particularly heavy, it floods the sewage networks, and the disease takes advantage of its enlarged habitat.

This is where this new computer software comes into play, a multidisciplinary marvel developed through considerable international coordination.

First, the UK’s Met Office gets a rainfall forecast for Yemen using its powerful supercomputers. NASA’s Earth-observing satellites also provide vital meteorological and climatological measurements.

This data is then plugged into a computer model developed by multiple US-based researchers, including hydrologist and civil engineer Antar Jutla at West Virginia University, and Rita Colwell and Anwar Huq, microbiologists from the University of Maryland.

It takes into account a dizzying array of factors, including population density, air and ocean temperatures, access to clean water, and even the measurements of phytoplankton concentrations along the coast. When given test runs in 2017, it proved to be 92 percent accurate.

Fergus McBean, a humanitarian adviser with the UK’s Department for International Development, spotted these results. At the start of 2018, he asked the team to help him, and others, develop a fully functional, real-time system for Yemen. They only had a time frame of 4 months to do so, as the rainy season was rapidly approaching.

Thanks to this system, scientists can find out where cholera cases will spike 28 days in advance. When predictions are made, aid workers then rush to those areas and begin vaccinations, treatments, and educational activities. This has already prevented tens of thousands of people from being infected with the potentially fatal disease. 

It’s worth noting that this is still a new model that’s being tested, and its precision may change from time to time. “The model has done an excellent job in Yemen detecting triggers of cholera outbreaks,” Jutla explained in a statement, “but there is still a lot of work we need to do to have this forecast model give accurate predictions everywhere.”

It’s best not to sign the cholera outbreak’s death warrant just yet, though. Cases have dropped considerably, but per The Japan Times, the UN is warning of a “third wave” of the epidemic. The rains will return, and the war shows no signs of abating.