Synthetic intelligence actually isn’t new to the digital well being area, however the sector appears to be within the midst of a brand new hype cycle. This one is concentrated on novel applied sciences like giant language fashions and different types of generative AI, stated Alex Lennox-Miller, lead analyst at CB Insights, throughout a Thursday webinar.
The healthcare AI house will likely be an thrilling one to look at over the subsequent couple years, as funding {dollars} move to startups and suppliers and payers launch extra AI pilots. Two of probably the most fascinating developments to look at would be the use circumstances that suppliers and payers will prioritize when deploying generative AI fashions, in addition to M&A exercise throughout the healthcare AI area, Lennox-Miller declared.
There are some areas of healthcare, like picture evaluation and digital pathology, which have been benefiting from AI for years now, he identified. However because the healthcare sector begins to focus extra on novel functions for AI, particularly generative AI, one of many use circumstances that’s getting probably the most consideration is automated documentation and ambient notetaking for suppliers.
Some firms constructing generative AI fashions for medical documentation embrace Nuance (acquired by Microsoft), Abridge, DeepScribe and Suki. It’s essential to notice that whereas this space is garnering consideration, the business’s efforts surrounding the usage of generative AI for medical documentation are nonetheless fairly nascent, Lennox-Miller stated.
“I believe we’ve seen some pilots which are very promising, however this can be a know-how space that we’ve seen individuals attempting to unravel for fairly some time. Whereas it’s fascinating to see the potential functions of generative AI right here — and there are some firms which have actually fascinating, efficient real-time merchandise — we’re additionally seeing a number of firms that also want a whole lot of work on their transcripts, documentation and guide corrections to supply actually usable documentation and notes,” he defined.
At the least for now, Lennox-Miller thinks that the usage of generative AI in healthcare will stay primarily on the “nonclinical aspect of issues.” He expects to see suppliers deploy these AI fashions to enhance affected person engagement, set up informational chatbots, summarize affected person histories for higher pre-visit planning, and ship sufferers residence with detailed plans for follow-up care. Generative AI additionally holds vital potential to automate referral letters and prior authorization requests, he added.
Lennox-Miller identified that 5 of the ten greatest digital well being funding rounds within the second quarter of this yr had been raised by AI firms. HeartFlow raked in $215 million, and Attempt Well being snagged $166 million. Moreover, Spring Well being, BenchSci and Flywheel all closed financing rounds totaling greater than $50 million.
Aside from Attempt Well being’s Sequence C fundraise, all these firms’ financing rounds happened on the Sequence D stage or later. When it comes time for later-stage healthcare AI firms like these to make an exit, Lennox-Miller predicted that the majority will likely be purchased as an alternative of submitting an preliminary public providing.
“We’re seeing a whole lot of consolidation within the house — a whole lot of acquisitions by bigger firms that want to add performance to an present platform. That is nearly reaching some extent the place our projections are hitting practically the extent of M&A that we noticed in 2021, on the peak of AI enthusiasm. However there has actually solely been one vital healthcare AI IPO, which was from Bullfrog AI in drug improvement. So whereas the house is getting a whole lot of funding, we’re additionally seeing a whole lot of consolidation and a whole lot of acquisition,” he defined.
In Lennox-Miller’s view, AI is an space of know-how through which giant incumbents have vital benefits over startups. Tech giants like Microsoft, Google and AWS have much more assets and computing capability, enabling them to successfully deal with the huge volumes of information wanted to develop a powerful AI mannequin, he stated.
“Frankly, AI is usually a matter of iteration. The extra you are able to do it and the extra time you may take to refine and construct it earlier than it has to grow to be a worthwhile product, the more practical will probably be. So M&A very represents these bigger organizations which have these pure benefits within the AI house profiting from what they see available on the market,” Lennox-Miller declared.
Picture: metamorworks, Getty Photographs