 Thanks for joining us at the AI for good Global Summit 2019 here in Geneva. I'm delighted to be joined by Thomas Vigand He's professor of information technology at the Berlin Institute of Technology an executive director of the Freinofer Heinrich Hertz Institute Thomas, thank you very much for joining us. Thank you for having me now Thomas You have been involved with the summit from the very start three years ago and last year you kickstarted an interesting project It's around the use of artificial intelligence for health What can you tell us about it? So the idea came out of the health track last year where 15 very interesting projects have been presented how AI could be used to improve health Have like diagnostic prediction or to have other predictive models But it was it became clear that When these projects are over, that's the end of it more or less and how do you get these AI algorithms to the field? How can they be used to a worldwide? Audience of two world-wide group of patients So Thomas out of the 13 projects one in particular is quite advanced isn't it? And it's one that uses artificial intelligence in the diagnosis of breast cancer. Is that right? Yes, so in Pathology when tissue samples are taken that needs to be understood where the tumor cells are and where the lymphocytes are They are the ones that infiltrate the tumor to fight it and it's important to know the quantity of those Currently a pathologist looks at those pictures and gives an estimate all counts and that takes time so an algorithm can do this much faster with a higher accuracy and The task of the focus group not only for pathology not only for diagnosis But for other topics on AI for health is to evaluate AI algorithms So we would go back to first the data the reference data, which need to be of high quality For instance in this history pathology case. We have Seven pathologists looking at the same picture to give us the gold standard as the reference and then we would be Testing AI algorithms against the gold standard and see how well they do and if they do good enough Then they can be used in the clinical Environment, okay, so you are trying to create a benchmark really. Yes, so Either it's a benchmark of the AI algorithm against the gold standard, which is the health experts or it is Among other AI algorithms and how well they are and then at the end we will draft a report and provide this information We have made great progress for instance We have created a working group on regulatory considerations with participation from the FDA of the United States and HPMA of China and EMA and Europe and the BFR AM of Germany and others are joining and so what we are doing is we are involving those Who are in the business of? regulating and certifying AI Into our work to give us guidance so that we can come up with this framework on benchmarking these algorithms So that at the end of the day, they could be made available to everybody who needs health improvements It's an interesting example of open innovation and collaboration as well, isn't it because you are working with so many different stakeholders. Yes So it is a group as an ITU focus group in collaboration with WHO But then there's the ENP involved the International Association of Nothing National Peptic Health Institutions the Interacademy Partnership which is the academic Academies in the world the ACM so we have to bring in all sorts of stakeholders and many many more in order to make this work because what it needs is many experts that are knowledgeable about health a particular health topic and about AI and statisticians so that we have a credible judgment over what we're doing is actually working and it can be used then on patients Fantastic. Well, Thomas. Thank you very much for talking to us today. Thank you. You're welcome