Bloomberg Audio Studios, Podcasts, radio news.
You're listening to Bloomberg Business Week with Carol Messer and Tim Stenebek on Bloomberg Radio. Okay, get this, Carol. In the US, more than twenty two point six million people work in healthcare a lot. It's about fourteen percent of the US working population. This is according to the Bureau of Labor Statistics. Yeah, healthcare pretty huge. There's a ton of money there and there are a lot of inefficiencies.
We hear about that.
You ever tried to go get an MRI like sent from one healthcare provider to another.
It's like impossible.
You got a cd ROM seriously, like way they still make that. Don't even get me started on faxes because yeah, doctors are still doing that too.
I know.
Well. Anterior is trying to make the system more efficient. It's an AI company. It's focused on healthcare administration. It's backed by Anya, Sequoia Capital and more. We've got with us doctor abdel Mamod, CEO of Interior. He joins us from London, Doctor Mahmood, good to have you with us this after this evening. Thanks for staying up late. We're in sort of the chain of the healthcare ecosystem, do you fit.
Yeah, well, lovely to meet you and thank you for having me on. So we tut of the main area is administration. So we were born to tackle the trillion dollars that I think the US spends on healthcare administration every single year. And administration is made up of many, many workflows, but at the core of it, and I think you mentioned this statistic, there's a lot of manual labor. There's a lot of very manual, tedious things involving factes,
and yes that's true. Doctors still use factors today. So we've started mainly on the health insurance because we feel like a lot of that administrative workflows and burdens starts from that side, and we focus there, and maybe they're the easy way to say it is we're focused on wherever there's a fax machine that a kind of a highly trained nurse or adoptor needs to spend hours reading, that's the bit that we make faster.
Okay, So I feel like folks have been trying to figure out how to modernize healthcare records and then also protect patient privacy. So tell us about how you guys are going to actually figure this one out. Why you versus everybody else who's been talking about it and trying it. Why you can do this.
Yes, that's a great question, and I think you touched on something important there, which is patient privacy and interoperability, right. And the reason we have facts today is actually because it's one of the most secure forms of getting two systems that don't know each other kind of aren't connected electronically to be able to communicate in a very safe way. The result in issue of that is that when you send over a fax, the system on the other side
receives the facts, but it can't ingest the data. Right. It's four hundred pages of medical information that only a human today can read, so you hire a doctor or
nurse to just read it. And I think the bit that's made us stand out in the market and we have the leading product in the market that we be able to understand unstructured medical data and we're almost like recreate the structured version of that and do interesting things with that downstream, such as bio authorizations, payment integrity, and
a few other certain workflows. And that's the the bit that's the most valuable here is you can maintain if you still use facts, and we should be getting rid of facts, but it remains a kind of a core part of of the US healthcare system. It allows you to kind of still have that digitized and still have that interoperable to that you know, of most valuable assets and resources nurses, doctors, pharmacists on not spending hours reviewing facts.
Is this for the US healthcare system only?
Yes? I think the US healthcare system is quite unique in the sense that it spends so much on these administrative workflows that are quite outdated in a sense.
Yeah.
Yeah, So where were you in your medical school journey or in your career where you discovered that the US has these inefficiencies. Take us to that moment.
Yeah. So I trained in the UK, and somewhere along the journey I realized that, you know, the impact you can have in healthcare is probably less about how better I got as a doctor clinically, but more about the environment and ecosystem and the tools that we use as condisions. So I embarked on a master's degree in computer science and spend some time actually Facebook and then later at Google working on some healthcare applications, and that's where I
got exposed as you know, those are American companies. I got exposed to the US healthcare system and I saw that, yes, there are issues of this in Europe, and there are issues in Canada and other places, but in the US it's really magnified and it's blown up in the last ten years or so. And that's what got me exposed. And that's when some of the large language model staff and GENERALLYVII that was coming out, felt like that was the most exciting thing you could apply large language models to,
which is this huge problem of unstructured data. That guess what, large language models inherently have that capability that we never had before of structuring unstructured data.
Go ahead, well, you know, it's interesting. I do think about AI and large language models and how they might be able to look at an X ray, you know, and compare it to all these others, you know, in a database and maybe figure out when there really is something wrong and do it rather quickly versus waiting for humans or maybe there's a doctor who's tired and it
doesn't come out so well. Having said that, I'm reading from the Sequoia website says that what you guys are doing handles all the work that comes before the clinical decision making, where it might take bless you a nurse several hours to track down, organize, and summarize hundreds of pages of records they need, Interior can do it in seconds, allowing clinicians to increase their output tenfold and focus on what matters most, helping patients love that love that mission.
How do we know that LLMS that AI is going to make the right clinical decisions?
That's a great question, and that's because you never let the large language model make the clinical decision. That is the key bit here, which is everything up until the clinical decision is the bit that takes the most time
and the most effort. So if you look at the nurses workflow right, they'll spend kind of of ninety five percent of the time logging into the twenty different systems right downloading the right PDF files, finding the clinical reference and documentation and guidelines, mapping it all out, finding needles in a haystack across hundreds of pages to then just have all the information gathered so that he or she
can make that decision right. So what we tell our nurses that use the product and the reason they love it is like each one of you gets an intern that overnight really smart intern that overnight has prepped all the nurs right, has gone through everything, and then lays it up for you. So you make the call, right, and that what that is is you ensure that nurses
and doctors go through administration much much faster. But they focus on the bit that they're great at the top of their license, the thing they went to medical school or nursing school for many years for.
So I want to know about training these models because there's a big issue when it comes to what this stuff is trained on. How do you do this when patient data is confidential?
Yeah, so flatter out. We don't absolutely don't train on any customer data, and I don't think companies would even allow you to, right, So that's all in the contract that we sign and so on. I think the world has moved on in a sense of the previous issue with AI was that you needed to train a lot of medical data to even have something viable. But I'm sure you've seen chat, gipt passes, the USMLI and all
these things. A lot of these general models that we use off the shelf right, they're actually pretty good clinically. The trick is that last ten percent though, And the reason why chat gipt isn't doing any of this stuff today is because it maybe gets sixty percent of the way there. How you orchestrate them, How you get multiple large language models, including also built on top of the decades of machine learning progress that we've had, on top
of databases and and other kind of computational techniques. How do you stitch that all together in end to end experience that feels like almost almost like a kind of completes the picture with a great user experience as well. That's the bit we focus on, and that's the bit that actually delivers them value.
Well, let us know how things are going. Doctor Abdel Mahmoud, Chief Executive Officer of Interior, joining US