umlaut daten worldmap

Interview

AIOLOS Predicting the next pandemic

Can AI help societies better prepare for pandemics? Caroline Weber and Max Haberstroh on umlaut’s new Franco-German research.

Covid-19 seemed to take everyone by surprise. Your new research project, AIOLOS, Artificial Intelligence Tools for Outbreak Detection and Response, could help to predict future pandemics. How does it work?

Caroline Weber: On a basic level, our aim is to collect data from many different sources and identify patterns that enable us to predict pandemics. The data could include mobility trends, public data and social media data, data from hospitals or medical research institutes. Ultimately, this could enable us to prevent another pandemic. Or at least minimise its impact by taking early measures.

Does anything like this exist already?

Max Haberstroh: Only partly. The idea itself isn’t new, but previous efforts have relied on either medical data or social data and didn´t combine the two. AILOS we not only combine data from many different sources but also bring together expertise from data science, medicine, and pharmacy – across academia and industry as well as from Germany and France. This willingness to match the complexity of the challenge is what gives us a real chance for success. It’s important to stress that this is a research project, though. There won’t be an industrial-ready tool at the end of it, but it could form the basis of one in the future.

Why hasn’t it been done before?

Max Haberstroh: For all the great things we hear about AI, people tend to forget that it starts with the availability of data. AIOLOS is just getting started and there are still a lot of question marks over exactly what kind of data we’ll be able to combine. Obviously, we have to consider things like data protection and privacy. But collaborating with such a broad cross-section of partners should help. The project is funded by the French and German governments, and we’re working with the Fraunhofer Institute for Translational Medicine and Pharmacology, CompuGroup Medical, an e-health company, Quinten Health, an expert in AI and medicine, and French pharma company Sanofi Pasteur. In addition, we already have a very interesting list of associated partners that expressed their interest in the project, and we are looking to get even more partners involved during the project.

What is umlaut providing?

Max Haberstroh: We lead the data engineering and the visualisation work stream and join forces with the partners on the analysis part. We’ll have to do a lot of data preparation and cleaning – taking data from hundreds of sources and adapting it into a single, comprehensive data set. In addition, as umlaut we do not only have proven expertise in working with sensitive data but also do have our own capabilities and infrastructure to collect and provide information that might be valuable for this use case. All being well, we aim to have some initial insights after the first year.

So, how easy is it to predict a pandemic?

Caroline Weber: It’s incredibly difficult if you don’t know exactly what you’re looking for. Take Covid-19: at the beginning, nobody would have known it was a new pandemic based purely on the symptom-related trends.

Max Haberstroh: Also, nobody is going to be able to just write an algorithm straight away that allows us to detect arbitrary pandemics. There will be an awful lot of trial-and-error work to identify any common threads that help us to spot emerging pandemics. Of course, we have a ‘live’ case with the corona pandemic right now, but it’s just one example. For this exact reason it so important that the project not only includes technology and data experts but also specialists from the medical and pharmaceutical field.

How does it feel to work on such a groundbreaking project?

Caroline Weber: It’s incredibly motivating. It’s not every day you get to do something that might save lives. We also should mention that we do have the honour to work on this project with great colleagues from different groups such as cloud specialists and data analysts within umlaut who will be crucial for a successful outcome of AIOLOS.

Max Haberstroh: Absolutely. I also really hope it shows that we can do a lot more with data science than just earn money. In my experience, there’s a huge gap between our capabilities in industry and those in the public sector. Projects like AIOLOS could help close the gap and bring a huge benefit to society.

Thanks for the interview!