The highly trained ‘eyes’ of Harrison AI - Jarrel Seah

And how does Harrison tackle that bottleneck, if we sort of summarize briefly what Harrison AI does?

You know, Harrison obviously doesn't fix everything, but one of the key problems in medical imaging is that not only is it hard to acquire the images, because there's, you know, million-dollar machines, there's very expensive labor and radiographers, but also the interpretation of these images is bottlenecked.

And if you look at the cost of a scan, somewhere around sort of like 40% to 60%, depending on which part of the world you're in, actually goes to paying the people who look at the scans, not the actual machine or not the electricity cost or not the contrast that we inject.

And that's how much you pay radiologists, right? And one of the problems is that, you know, we always have to look at every image in every study, right? And as time goes on, the scanners get better.

We get more and more images.

We get different types of images and the scans become more complicated and harder to interpret.

So not only are we getting more scans, we're also getting scans that are harder to read.

And so Harrison wants to make that process easier.

We want to provide a second set of eyes for physicians and surgeons and other doctors to look at the images and be confident that they're getting the right diagnosis or the right findings and improve efficiency across the board, essentially.

Yeah, amazing.

And your role at Harrison is both chief medical officer and chief AI officer.

And you don't hear those two together that often.

Yeah, no, that's really unusual.

That sounds very cool.

So Harrison, the way it interprets the scans is very much driven by AI and training, but perhaps you can sort of talk to your role in that and how AI is involved.

Yeah.

Harrison AI, as the name implies, is primarily an AI company.

We train AI models to read these images and identify the types of findings that a radiologist or other doctors might care about.

Put simply, we want to provide comprehensive AI solutions where other AI products on the We historically have provided a single diagnosis or a single finding.

So you might be able to find, say, quite commonly an AI that might detect a pneumothorax, which is sort of a punctured lung on an x-ray.

That's not just the only thing that we look for on the x-ray though, right? You can look at fractures, you can look at cancers, you can look at bowel problems, heart problems.

The unique thing about Harrison is that we want to do everything.

We want to approach it the way that a doctor might.

When Harrison was started, it was quite clear that the way that the products had to be built had to be built by somebody who had both AI knowledge as well as medical knowledge.

So we approached this very similarly to how a doctor in training might do it.