AI Investor Forum: Gus Tai on AI in Healthcare (Part 3)

Sramana Mitra: Now, let’s shift to medical imaging, which is another area where AI is having lots of impact. There are lots of companies working in this field, and lots of entrepreneurial efforts going on in this field. The question that I’m of thinking about is that, are these going to be unicorn style opportunities?

The million pound gorilla in healthcare records is Epic, which is in almost every hospital system. Epic happens to be a private company. What is going to happen? It has enormous amounts of data. Almost all this data goes through Epic, which has a very large market share.

So, what kind of a market dynamic are we getting into here?

Somebody entrepreneurial has an AI software to do some kind of analysis; then as the company gets some validation, companies like Epic or other companies in that class just buy these companies? What are we looking at? Are these gonna be independent companies? Are these all going to be relatively bootstrapped to exit or capital efficient exit focused companies?

What kind of a market dynamic are we looking at?

Gus Tai: You’re raising two interrelated topics. One is around diagnostics capability, and the other is the central storage and distribution of data and information.

My understanding of Epic, a private company that I’ve known since the nineties, is that they don’t specialize in advanced diagnostics. Instead, they serve as the lifeblood or sales force of a hospital and healthcare system, focusing on taking in data or information feeds, making it intelligent, and redistributable.

As an aside, there’s a lot of activity in AI for ingesting data for Epic and other services. Epic itself offers a service that can record a patient visit, take in the transcription, and add metadata to it. There are also startups funded to do that.

On the diagnostic side, I believe there are startups that could be bootstrapped or require low capital. Once again, my familiarity is with the US healthcare system. In the US healthcare system, there are rarer disease types that create significant concern for patients and the medical community. AI could be applied to reexamine an area, leading to better analysis and conclusions through domain expertise and appropriate AI algorithms.

This leadership position could attract people willing to pay for better diagnostics, which is essentially a painkiller or therapeutic. This area of expertise might appear small initially but could become very large. It was like the company that you’ve worked with or you know quite well of the pharmaceutical SaaS company. I’m forgetting the name.

Sramana Mitra: Veeva.

Gus Tai: At first when it started, they’re like, “well, how big can it be?” Well, it’s what now? A couple of hundred billion? I mean, it’s an enormous company.

Sramana Mitra: Veeva is an enormous company. Veeva started with being the CRM for the pharma industry. I do think in medical imaging, you can build very large companies. That’s the reason I brought up the bootstrapping issue.

The vast majority of the industry prefers to see validation before they’re willing to write checks. We’ve talked about this before. The comfort zone of investors is when there is some product, some product market fit, or some validation, then they’re willing to write the checks.

Now, the world in which we operate, there are so many people who are getting to that valid, who are trying to get to that validation, trying to be able to become fundable. How do we bridge that gap in areas that are inherently capital intensive? This is why I’m starting with this discussion.

If you want to do a lot of different kinds of medical imaging or diagnostics kinds of specialized diagnostics types of companies, what is the path to proof of concept so that they can then access capital? I think they can be big companies and they can access capital, but that proof point period, the MVP and early product market fit period is still very tricky.

Gus Tai: To your point, Sramana, what comes up for me is that your audience has a large community of international participants. For some of these rare diseases, which are human diseases, they occur everywhere around the world. It’s a matter of validating the testing of the algorithm. It might be easier to aggregate the data and more affordable to test overseas, and then bring it to the US.

I think that’s pretty interesting. This would involve determining where you can access the data, where you can perform the tests, and if you find significant improvement in the analysis, that would be valuable IP everywhere in the world.

This segment is part 3 in the series : AI Investor Forum: Gus Tai on AI in Healthcare
1 2 3 4 5 6 7 8