Can we fight malaria with AI?

Completing an escape room in Nijmegen with top Google researchers. Who can say they did that? Koen Dechering got AI help from Google’s computational scientists in his studies on tropical infectious diseases.

Koen is the CEO of TropIQ Health Sciences. He helps organisations to discover drugs and vaccines for tropical infectious diseases. They have a long-standing collaboration with the Bill & Melinda Gates Foundation, known for its work in fighting poverty. “The foundation put us in touch with Google’s top researchers”, Koen explains. “They were keen to run a few projects to show how Artificial Intelligence (AI) can impact poverty alleviation and reducing disease burden.”

Malaria never went away

Research into tropical diseases is more urgent than ever. “You could say we are facing three simultaneous pandemics: malaria, tuberculosis and HIV. They received less attention during the Covid pandemic. But they never went away. Precisely because the care for these diseases was downgraded at the time, it is now of enormous importance. Our goal is to improve people’s opportunities around the world by improving their health. Health and economic status are connected: if you are poor, you’re more likely to get sick. And if you’re sick, you can’t participate in the economy. Something has to be done about this vicious circle.” “So the call from Google came at the right time. We thought the combination of AI and scientific research could help us reach our common goal: helping our global health system in discovering novel mosquito-repellent molecules. Now is the time to take action.”

Let’s Go(oogle)!

“What did we do? Google explored the possibility of giving computers the ability to smell and trained computers to recognise specific scents. More specifically, to predict the scent of molecules based on their structure. Put in a molecular structure, and the computer will predict whether it smells like lavender, vanillin, or any other scent. It's all about repetition; think of it as learning a new language.” The next step was to insert the most important cause of malaria: mosquitoes. “Smell is their most important sense. So, we immediately linked the technology to mosquitoes. They smell blood and choose their next prey based on that smell. We found an old dataset, boasting roughly 19.000 measurements of mosquito repellency. We discovered mosquitos respond to scent molecules by feeding this data into the Google model. If you know what smells mosquitoes like and which ones repel them, you discover powerful tools to stop malaria.” The results were revolutionary. “We tested the AI against older methods, and AI did way better. It was like having a smart assistant who could quickly figure out the best mosquito repellents by simply smelling different scents. We even found more than ten new molecule formulas with repellency that work as well as, or even better than, the ones we already have. We now know which scents repel mosquitoes, which is essential knowledge in our fight against malaria.”

The future of AI in the healthcare business

“One thing this study has proven is that the process is way more efficient when using AI.” So, should healthcare businesses embrace AI for its innovative and efficient solutions to complex challenges? For Koen, it’s a big yes. “We want to use our experience with Google in other upcoming projects. For example, we are also interested in substances that kill ticks. We are now training AI models to find those substances faster. That way, we can better protect people from Lyme disease. Because in the end, our goal is always the same: to improve people’s chances by improving their health.”