Healing Healthcare with Artificial Intelligence : ‘Beyond The Hype’ with Dr. Ali Parsa, Founder and CEO of Babylon Health

March 29, 2018

Written by


MMC Ventures Podcast


How can AI put accessible, affordable healthcare into the hands of every person on Earth? We go ‘Beyond The Hype’ with Dr. Ali Parsa, Founder and CEO of Babylon Health.


As Babylon develops cutting-edge AI for medical diagnosis, Ali describes the opportunities, challenges and implications of reinventing a $10 trillion industry using AI.


With Babylon valued at over $200m, Ali also provides an entrepreneur’s perspective on how to build a world-class AI company. How can organisations recruit top AI talent?


How can startups sell to the NHS? What personal characteristics are key for entrepreneurial success?


Click here to listen to the Podcast - https://www.mmcventures.com/podcast/episode-2-healing-healthcare/




Highlights from the Podcast


What AI has Babylon built?


Let me explain to you what we have built, because people talk about machine learning, but there are di erent branches of course and techniques in arti cial intelligence. A better way of thinking about this is: what is it that we create? If we create an arti cial intelligent doctor, we’re basically replicating the brain of a doctor. So, what does a brain of a doctor contain? We send a student to university and they learn. We give them a knowledgebase. So, we have created in here what is today probably the world’s largest knowledge base in medicine, certainly in primary care, with over ve hundred and thirty million strings of knowledge. No human brain can contain that many, or one that is at least accessible immediately.


Second, a doctor talks to you, interacts with you, to check, take information from you and to convey it to you. So, we have created a natural language processing, understanding, generation capability here that can understand medicine. So, for instance, when I say my tummy is killing me, it understands it means my stomach hurts and it doesn’t call the police.


The third thing is a doctor then collects that information and infers from it. It logically decides what it is that is wrong with you. For that we had to create what is today, I understand, the world’s largest Bayesian network - the largest probabilistic graphical model that ever existed. And it’s natural that it should be there because you look at the combinations of symptoms and risks and history you have. It’s billions of variations that we’re trying to assess to set the probability of what disease you can have. We had to build that. And we had to build in a Bayesian network and not, as they call, in deep learning, because it needs to be auditable. I cannot say that ‘hey, the machine said that’s your disease and I can’t know why’. So, it had to use a particular technique as opposed to another.


And then a doctor says okay, so that’s your disease, and let me now tell you: if you don’t x it
here what is going to happen in the future. Or it’s what we call in arti cial intelligence, predictive analytics. And then, of course, a doctor every time learns from their interactions with a patient. That’s why we call it practice. And that’s the part where the machine learning comes. So, the machine learns from every interaction it has. 


Why do Babylon use Bayesian networks? 


In deep learning, you need to have enough data sets, reliable data sets, for the machine to learn. And in medicine, you know when the great team in DeepMind won the game Go against a world champion they trained the machine through a very large selection of games that the machine played. But the rules of the games are written in two paragraphs, give or take, for the game Go.


The rules of medicine are not even written in two volumes of books. And there is not enough data in the world to teach a machine just sort of randomly how to become a doctor. You need to create that knowledge and you need to teach the machine. So, there are other reasons for using the techniques we use, and we do use a lot of deep learning. But you use it in other parts of our business - we use it language, we use in the interactions, we use in other parts.


How do you think AI will impact healthcare overall? 



I think it will change everything. We are on the verge of being able, as I said at the very beginning of this conversation, to put healthcare into the hands of every human being on earth.


Never on the face of the planet has that planned. And we’re not going to put ‘a’ healthcare in the hands of everybody on earth. We’re going to put eventually the best available healthcare. We’re going to do with healthcare what I said Google did with information. Because it doesn’t matter how rich or poor I am. It doesn’t matter whether I am in Barcoo or in Baltimore, in Kabul or in Connecticut. I have access to the same information through Google.


That is just not true about healthcare. The rich have much better healthcare today than the poor. It’s not even the same league of healthcare they have. And I think we can equalise. We can truly democratise that. We can make it possible for a peasant in India to have the same access to diagnosis to start with, and healthcare eventually, that the rich in Washington or in California have. 


Click here to listen to the Podcast - https://www.mmcventures.com/podcast/episode-2-healing-healthcare/




About MMC Ventures


Founded in 2000, MMC Ventures is a research-driven venture capital rm investing in high-growth technology businesses. MMC has backed more than 50 enterprise so ware and consumer internet companies that have the potential to change the future of nancial services, the workplace and retail.


MMC invests on behalf of private and institutional investors. MMC has over £200 million under management and is investing approximately £35 million annually.


MMC’s portfolio includes: Appear Here, Bloom & Wild, CloudSense, DigitalGenius, Elder, Gousto, Interactive Investor, Masabi, NewVoiceMedia, Safeguard, Senseye, Signal Media, Sky-Futures, StoryStream and Tyres on the Drive. 


Source : https://www.mmcventures.com/wp-content/uploads/2018/03/Babylon-transcript-v4-1.pdf



Share on Facebook
Share on Twitter
Please reload

Please reload