In 2025, the healthcare industry will expect more out of artificial intelligence.
Vendors will push AI further, healthcare delivery systems will hold AI to higher account and the line between telehealth and AI will blur.
This was the resounding perspective from health tech founders that are working to stay on the cutting edge of healthcare.
If 2024 was the dawn of the ambient AI scribe and a gully for the telehealth industry, 2025 will bring AI quality to bear, assisting virtual providers and in-person providers alike.
“One year ago, when we did a demo of our product, clinicians were really amazed,” said Alex Lebrun, co-founder of Nabla, which offers an ambient scribe tool. “They thought it was pure magic … fast-forward 12 months, we don't even need to do demos. They know everything about it … they have more precise questions, expectations about accuracy, how the product will adapt to a specific workflow or clinical specialty, how it will integrate with [their] EHR.”
Julia Chou, chief operating officer at Abridge, which also sells an AI scribe, said the healthcare industry’s understanding of AI scribes grew exponentially from the first quarter to the fourth quarter this year.
“As everybody is getting more up to speed with this technology, how to deploy it in their systems, I think there's a lot more focus on, is there true science behind this? How are you actually evaluating the accuracy and the efficacy of the algorithms that power things? What are the actual integrations that enable a seamless user experience on the clinical side?” Chou said.
Given the copious attention on AI scribes this year, healthcare providers' expectations for the technology are rising, founders and executives said.
Healthcare providers who have used an AI scribe solution in the last year will expect more accuracy from their scribes and will expect them to pick up on language that is specific to their specialty. Providers will also gravitate to solutions with more customizability, LeBrun said. The quality of the output, the level of integration with the electronic health record and the workflow will separate wheat from chaff in 2025.
“A generic ChatGPT wrapper won’t make the cut anymore,” LeBrun said, referring to AI scribe products that have been quickly built on foundation models like OpenAI’s ChatGPT. These types of wrappers are easily replicable and are susceptible to external changes.
Mudit Garg, CEO and co-founder of Qventus, an AI operational assistant company, explained why AI has to improve.
“The audience is already so burnt out, the physicians, and they have such low tolerance for it not working,” Garg said.
All of the executives interviewed expect the AI scribe market to consolidate next year. From the roughly 60 AI scribe companies that are currently on the market, Garg predicted the market would shrink down to six or seven AI scribe companies.
Two founders predicted that by the end of 2025, roughly 30% of the healthcare market will use an ambient scribe, though the usage won’t be evenly distributed through care settings. Garg predicted that 40% to 50% of large academic health systems might be trying some version of an AI scribe by the end of next year. Outside of academic centers, adoption will hover between 15% to 20% of clinicians, he predicted.
Punit Soni, founder of Suki AI, predicted that there will be only two or three AI scribe winners by the end of 2025. The winners, all interviewees said, will be the companies that iterate beyond ambient scribe to an AI assistant. The companies that survive will also be those that do hands-on integration and problem-solving with their customers, multiple founders said.
Soni expects 10% of the healthcare delivery market to adopt an AI assistant by next year.
All of the AI scribe companies Fierce Healthcare interviewed said they are striving to become AI assistants rather than just AI scribes.
As AI vendors build out the capabilities of their tools to be more useful to clinicians, healthcare purchasers are getting over the fanfare of AI and have executives reviewing the brass tacks.
“What I'm seeing these few last months is finally the CFO wants to see the return, and I think it's a big change coming that we need to prove the hardcore return on investment of the AI tools,” LeBrun said.
“Trust and auditability will continue to be key driving factors when evaluating the efficacy of systems and also a stronger focus on science,” Chou said. “I think that we're seeing that health systems are starting to bring in AI officers, for example, and they're really looking for strong science-backed research teams that can partner with them to develop the next set of technologies.”
Garg said 2025 could be a year that physicians discern what they want to use AI for and what they don’t.
Companies like Dandelion Health are trying to provide assurance about the use of clinical AI and help providers determine where AI is most useful. Elliott Green, founder and CEO of the real-world data platform, said healthcare will explore low-risk clinical applications of AI as well such as faster diagnosis and triaging and otherwise augmenting providers.
“In 2025, there's probably going to be a moment of actually trying to find these examples, very strong examples, where you know, everyone kind of wins using AI,” Green said. “What are the disease states? What are the tests? What are the AI algorithms where the payer wins, the patient wins, and also the health system wins?”
Virtual care 'ripe for AI'
Telehealth companies are also leaning into AI to propel them forward in 2025. Executives see the potential for AI and other digital health technologies to transform the U.S. healthcare experience. The distinction between the technologies will blur, telehealth executives told Fierce Healthcare.
“Virtual care is ripe for AI,” said Michelle Davey, founder and CEO of telehealth engine Wheel.
Geoffrey Rutledge, M.D., HealthTap co-founder and chief medical officer, pointed to HealthTap’s current use of AI, a chatbot that takes in patient information and asks relevant follow-up questions before the visit.
Rutledge said the application of AI is a value-add because it is doing more than summarizing known information. The model collects additional information to maximize the time clinicians spend with the patient during the visit.
“The big winners are going to be people who have the infrastructure layer of virtual care delivery, because we're the ones who can push the innovation the most,” Davey said. “If you don't actually own the EHR or own the experience, you're really innovating at the fringes of care delivery versus down into the actual guts of care delivery, which is where the most impact is going to be made.”
Davey and Rutledge said telehealth companies will amass power by integrating point solutions into their broader platforms, in the so-called “platformization” of virtual care. Moreover, Davey and Rutledge believe telehealth has iterated since it was brought to the fore during the COVID-19 public health emergency.
“[There has been a] realization that virtual care can fill in the gap for access to primary care, when you think in particular, Medicare and Medicaid populations have restricted or difficult access to care, and I see a huge value of telehealth delivered to that patient population,” Rutledge explained.
Rutledge pointed to some of what he sees as the failures in telehealth so far, like prominent retailers entering the space and standalone direct-to-consumer telehealth platforms whose overhead costs exceed telehealth reimbursement rates. Despite these “failures,” telehealth is proven and ready to enter a transformative phase.
Roy Schoenberg, M.D., Amwell executive vice chair and CEO, said that telehealth will be a sort of Trojan horse for other healthcare innovations and software-based treatments. Like Davey, Schoenberg said the companies with the platform power are going to win out and house the innovations of the future.
Schoenberg thinks there is an opportunity in the incoming Trump administration to make some of the changes that need to be made to rethink the cost savings of digital health.
“We know that we could move the needle on the GDP,” Schoenberg said. “I'm not even talking about Medicare or Medicaid. If we allowed elder Americans to spend more of their elder days at home, rather than in a skilled nursing facility or senior living or inpatient facilities.”
He continued: “These are enormous numbers that taxpayers have to essentially take and pay. if we use technology to support how people can stay home longer, through AI, through telehealth, through biometrics, through RPM, through fall detection, a variety of other technologies. If you brought these together, we could fundamentally change our healthcare expenditure.”
“In order to really envision this happening you're gonna have to kick a lot of sacred cows,” Schoenberg said.
While some of these changes will take years to realize, 2025 is likely to bring more innovation in AI and virtual care.
“I think I'm the most excited I've been in the last 13 or 14 years about what's coming,” Garg said.