One of the most promising areas of research with AI involves applications to healthcare. This is a field where data is ultra-important, and artificial intelligence has a lot of potential to innovate and bring solutions to the table. So in the race to find the best use cases, there’s a general consensus that the insights and pattern recognition that AI enables can help clinicians a lot.
But how does this work?
“The application of technology and artificial intelligence in healthcare has the potential to address some of these supply-and-demand challenges,” writes a team of authors of a paper published at the NIH National Library of Medicine. “The increasing availability of multi-modal data (genomics, economic, demographic, clinical and phenotypic) coupled with technology innovations in mobile, internet of things (IoT), computing power and data security herald a moment of convergence between healthcare and technology to fundamentally transform models of healthcare delivery through AI-augmented healthcare systems.”
Certainly, many of these applications are related to genomics. Others have to do with bringing data local on wearables for clinical analysis.
What all of these have in common is that they need the right platform and infrastructure to be effective.
Talking at Imagination in Action
(Disclaimer: I help to put on events for Imagination in Action, with emphasis on the MIT community in Boston).
At this year’s April conference, we had a panel on AI and healthcare infrastructure, where Karuana Gatimu of Microsoft interviewed Aditya Sharma of Stanford, Sufian Chowdhury of Kinetik, and Brad Reimer of Sanford Health.
Starting off, Gatimu asked Chowdhury if, prior to AI integration, non-emergency medical transport as a service was already “broken.”
Chowdhury explained how his business evolved to fit changing needs.
“A decade ago, when I started the company, I started it off as a revenue cycle management platform,” he said.
Then the idea evolved.
“We decided to build this end-to-end tech stack that allowed health plans to adopt our booking platform, and then we integrated into the computer aided dispatch systems that local transport companies were using, and then we’d fetch that data back to our RCM platform that was connected to all the payers clearing houses,” he said, explaining how data transparency works for rehab and dialysis patients. “And so that allowed us to create a closed loop solution.”
Listening Along
Reimer talked about a technology called “ambient listening,” which is when AI listens along to a conversation between a doctor and a patient, records it, and then brings out actionable data from something fundamentally unstructured. This also saves the doctor the work of taking notes.
“There are improvements in the quality of that data,” Reimer said. “It does improve the patient experience and the provider experience, bringing joy back into medicine. A lot of physicians have experienced burning out just because of the amount of documentation that they were having to do. But what we’re finding now is that even that’s becoming kind of table stakes, as it’s not a new technology anymore.”
With the right privacy controls, you could see this kind of technology adding a lot to the healthcare experience.
Chowdhury added more about how AI can help humans to make an end run around laborious back-end processes.
“I think now more than ever, we can leverage AI in a way where we meet everybody where they are,” he said. “We take out a lot of the back-end complexities, and we don’t necessarily have to create an interface where folks have to interact with it.”
That sounds, to me, a lot like the ambient listening, where a doctor no longer has to do the note-taking. Chowdhury explained how this sort of thing used to work.
“Traditional SaaS models were: I’ll build out a SaaS product, and then I’ll go to you, whether it’s a health system or a health plan, and I’m going to teach you how to use the system,” he said. “And then for the members, I’m going to teach you how to use the mobile version of our platform. And so there’s this education process where you lose a lot. There’s a ton of attrition that takes place, simply because learning a new software can be overwhelming, especially when there’s a new type of software for every process within a health system or within a health plan.”
Of the company’s own process, he said this:
“For us, when we launch our end to end product, historically, what we do is we take our platform, we design call center around that, so the health plan could build out their call center, they learn to use our technology, and then they have hundreds, if not thousands, of local clinics and providers that are in their counties, that we have to then go stand those products out, train them on it, and then we have to go to the members and teach them how to use our mobile app,” he said. “Today, we could literally go to health systems and say, ‘here’s a number. Just call it,’ and then it goes into our agentic layer, which will be able to speak to those folks that could just use their telephone. So we could actually go back to the 90s, ‘use your phone and make a phone call.’”
Dealing with Privacy Issues
“Healthcare is full of walls,” Sharma said. “Part of it is to protect me and you. At the end of the day, we’re all patients, right? You don’t want your information out there.”
“We have all HIPAA compliant infrastructure,” he said of his firm’s setup. “We’re on all the clouds. We have all HIPAA compliant LLMs, and we triage all of the data. We’re doing crazy parlor tricks with the data, to be able to answer the question in low-latency format, with respect to the kind of walls that do exist.”
Sharma summed up some of the related patient privacy concerns this way:
“The data is data, yes, and it’s kind of decoupled from the patient, but it is the patient, right?” he said. “It’s their data. They have the right to manage their own privacy, to not have that just everywhere that it could be. So our team is very mindful of the privilege that we have of all of this data, but we also are very mindful that at the end of the day, it belongs to someone else.”
Reimer talked about data living in silos.
“We have the same challenges as a metro hospital would, because there are great solutions that are coming out, but each solution fits this one niche, and it doesn’t necessarily talk across the whole health system, or that whole customer continuum, or patient continuum,” he said. “I think the data liquidity and the ability to have that data tied across that whole experience for the patient is something we haven’t quite solved for yet, and it’s really important as we think about the future of healthcare.”
What’s Next
In terms of other utilities, Chowdhury gave some comments on what else we will soon see.
“We can help markets like home care, which we’re seeing greater adoption of across the healthcare ecosystem, where you can leverage the folks that are closest to you, like your parents or your children or your friends, to help you get the care you need, especially in rural areas,” he said. “So with home care, naturally for us to be able to digitally see what you’re doing, we need things like remote patient monitoring systems to go to you, right? So if you have a logistics infrastructure, it doesn’t necessarily have to be taking members to a point of care, which is, traditionally, you know, a clinic or provider setting, you can actually bring that care home, or you could also improve the utilization of telehealth.”
He had another point on telehealth.
“Telehealth, over the last decade or so, hasn’t lived up to its promise, simply because folks weren’t able to get the RPM products that supplement telehealth and virtual care to the homes of members,” he added. “So I think there’s just massive opportunity.”
I thought all of this helps to set the stage for innovations in the healthcare sector. What do you think? Drop a comment and let me know.
