- by Alison Booth

Artificial Intelligence: predicting dengue outbreaks and outcomes

 Image of a mosquito over a SIM card with text reading, 'Artificial Intelligence predicting dengue outbreaks and outcomes'.

Technology is becoming increasingly important in the fight against dengue. Our own Dengue Track initiative is translating data from a variety of sources to create useful alerts when people may be at risk. Under the covers, it uses big data methodologies to generate its meaningful information. Other emerging technologies, including Artificial Intelligence (AI), may also soon also play a critical role in combatting the disease. Let’s explore AI and dengue: how smart machines are taking the fight to this deadly disease.

AI is based on the idea that computers can perform tasks that require some intelligence – that need them to be ‘smart’, in other words. These days, AI tends to be about creating machines that mimic human intelligence: how we learn, carry out tasks and make decisions.

Machine learning is a specific type of AI that mimics how humans learn. Give machine-learning computers relevant data, tell them about it and they will acquire new knowledge. These smart computers apply step-by-step instructions known as ‘algorithms’ to the data to create a digital ‘model’ of the real-world problem they are trying to solve. They reuse this model to solve similar problems time and again, improving it with experience.

Back in 2016, we asked whether AI was the key to dengue prevention. Today, AI technologies are being used to help smart machines learn how to predict dengue outbreaks, predict how individual dengue cases might develop and guide drug discovery.

Let’s start by looking at how AI is helping to predict dengue outbreaks.

Artificial Intelligence predicting dengue outbreaks

When it comes to Artificial Intelligence predicting dengue outbreaks, it begins with the smart computer digesting and analyzing vast amounts of relevant data, about the environment and patient outcomes in dengue-endemic regions, for example. The computer then deciphers and learns which conditions are most likely to result in an outbreak, and from that estimates the dengue risk for any particular local scenario, as we can see in these examples:

Aime Inc has developed a tool for providing real-time predictions on the locations and timing of dengue outbreaks. The result of 2.5 years of research, it worked with “up to 84% accuracy” in trials in Penang. Aime now has contracts in Rio de Janeiro, Manila and Kuala Lumpur.

“The predictive platform can warn of possible outbreaks three months in advance and pinpoint them to within a 400m radius.”

• A study in China developed a model based on climate factors (including mean temperature, relative humidity, and rainfall), search query data from China’s Baidu internet search engine, dengue case data and human population data to develop an AI dengue prediction model.

“The epidemics during the last 12 weeks and the peak of the 2014 large outbreak were accurately forecasted.”

• A team in Manila developed a model that can predict the number of cases per 1,000 residents at locations throughout the city based on land use and rainfall levels.

“The computer learns over many iterations how to fine-tune its model to predict dengue occurrence from local conditions with increasing accuracy.”

• A study in India is looking to design a new model by combining various diverse machine learning models.

“As there are some prediction models are developed for dengue fever predication. The discipline is in its infancy and much work has to be done, so more research work is needed.”

Researchers in Colombia are exploring using data about the pattern of dengue and chikungunya outbreaks to develop machine-learning models to predict disease spread.

“The development of computational models to forecast the number of cases based on available epidemiological data would benefit public surveillance systems to take effective actions regarding the prevention and mitigation of these events.”

AI predicting dengue outcomes

One second possible way machine learning could help the fight against dengue is by predicting which dengue cases are likely to evolve into something more severe. Here, researchers have been looking for any genes that might mean a patient is at higher risk.

Read more on, ‘Advances in using Internet searches to track dengue

To build an AI model to estimate patients’ risk of ultimately developing severe dengue, a study by Stanford University in California fed their computer with vast volumes of health records from patients who had contracted dengue. They combined this with data on gene expression from the US National Center for Biotechnology Information (information on our how bodies convert DNA instructions into something they can use to fight the infection).

While the scientists have identified several important genes and their parts (gene features) linked to the risk of developing severe dengue, they still have more work to do:

“In a real-world diagnostic situation, the relative proportion of patients who will develop severe dengue would be much lower than in our study. […] The machine learning algorithms will likely need tweaking before deployment as a diagnostic tool.”

AI and dengue drug discovery

Our third example of AI helping the fight against dengue is all about treating dengue. Scientists have turned to machine learning to help in the early stages of developing an antiviral drug for treating dengue infection.

During its lifecycle, the dengue virus produces a single protein that goes on to create all the viral proteins it needs to do its work inside an infected cell, according to an article in Atlas of Science. Before they can look for potential new treatments to combat this, the scientists needed to identify where new drugs could attach to these proteins.

After building a machine-learning algorithm and feeding it with published experimental data and molecule descriptions they were able to “characterize novel sites in the protein complex as susceptible targets or anchor points for drugs against all four serotypes of dengue virus.”

In other words, they used machine learning to find new places for potential future dengue treatments to latch onto and impede the dengue virus, stopping it in its tracks.

With so many stories of AI helping the fight against dengue, we’d love to hear your stories of how you are engaging modern technologies in your own battle against the disease.

Help reduce the impact of dengue in your area. Click below to create a record of dengue in your neighborhood.

Dengue Track