The lessons from machine learning and AI
A few weeks ago we reported on how clinic chain Fullerton in Singapore has managed to cut the costs for providing healthcare for employees in large accounts by 10-15% by deploying machine learning/AI to scan big data sets. This exercise revealed huge wastage and fraud. Fullerton also claims its program can predict medical outcomes far more accurately than falible human doctors. So who else is doing this stuff today?
Oddly enough, the answer would appear to be almost nobody. Zeeshan Syed, a Stanford professor of AI and CEO of US start-up Health at Scale, says that many such exercises have been proven to work in lab conditions. But that rolling it out into the real world is much tougher.
That reflects several factors. He says that big data sets in healthcare remain incomplete. More to the point, the expertise to really interrogate them is often lacking. You need a team of doctors to interpret the stuff. We will publish a full interview with Syed shortly.
We think you also can not underestimate the resistance to the deployment of AI and machine learning in the public sector and many areas of the medical profession.
We also think the obvious place where this stuff will take off is Israel. Much of its healthcare is provided by highly efficient, vertically integrated payor/providers. So the many vibrant Israeli tech starts up have access to impressively complete data sets. If early man emerged from East Africa, AI/machine learning in healthcare will emerge from Israel.
The other lesson? We think that the winners will be for-profit operators who build service provision around a set of AI/machine learning capabilities. That describes babylon and a host of others. Introducing AI and machine learning systems that force change in existing provision will prove much tougher!
We would welcome your thoughts on this story. Email your views to Max Hotopf or call 0207 183 3779.