GenAI Still Working Toward Prime Time With Patients

When it rains it pours for AI research, and a trio of studies published just last week suggest that many new generative AI tools might not be ready for prime time with patients.

The research that grabbed the most headlines came out of UCSD, finding that GenAI-drafted replies to patient messages led to more compassionate responses, but didn’t cut down on overall messaging time.

  • Although GenAI reduced the time physicians spent writing replies by 6%, that was more than offset by a 22% increase in read time, while also increasing average reply lengths by 18%.
  • Some of the physicians were also put off by the “overly nice” tone of the GenAI message drafts, and recommended that future research look into “how much empathy is too much empathy” from the patient perspective.

Another study in Lancet Digital Health showed that GPT-4 can effectively generate replies to health questions from cancer patients… as well as replies that might kill them.

  • Mass General Brigham researchers had six radiation oncologists review GPT-4’s responses to simulated questions from cancer patients for 100 scenarios, finding that 58% of its replies were acceptable to send to patients without any editing, 7% could lead to severe harm, and one was potentially lethal.
  • The verdict? Generative AI has the potential to reduce workloads, but it’s still essential to “keep doctors in the loop.”

A team at Mount Sinai took a different path to a similar conclusion, finding that four popular GenAI models have a long way to go until they’re better than humans at matching medical issues to the correct diagnostic codes.

  • After having GPT-3.5, GPT-4, Gemini Pro, and Llama2-70b analyze and code 27,000 unique diagnoses, GPT-4 came out on top in terms of exact matches, achieving an uninspiring accuracy of 49.8%.

The Takeaway

While it isn’t exactly earth-shattering news that GenAI still has room to improve, the underlying theme with each of these studies is more that its impact is far from black and white. GenAI is rarely completely right or completely wrong, and although there’s no doubt we’ll get to the point where it’s working its magic without as many tradeoffs, this research confirms that we’re definitely not there yet.

Telehealth Linked to Quality as Extension Deadline Looms

Efforts to extend regulatory flexibilities for virtual care gained some wind in their sails from a new study in Health Affairs that linked telehealth use to significant quality improvements and a relatively modest bump in spending. 

Researchers assigned Medicare patients to health systems according to care patterns in 2019, then segmented the providers based on their telehealth adoption in 2020 (5.5M patients, 576 systems). They then analyzed outcomes for 2021 and 2022.

Patients at health systems in the highest quartile of telehealth use (27% of all visits) had an increase of 0.21 outpatient visits per patient per year, a relative increase of 2.2% compared to systems in the lowest quartile of telehealth use (9.5% of all visits).

  • These patients also had 14.4 fewer non-COVID ED visits per 1,000 patients per year, a 2.7% relative decrease.
  • Further, the patients at high-telehealth systems saw improved adherence to medications like metformin and statins, although there were no clear changes in hospitalizations.

Those improvements came at the cost of an additional $248 per patient per year at high-telehealth systems, a relative increase of 1.6% above the lowest quartile.

  • The authors noted that this increase was largely driven by inpatient admissions and pharmaceuticals, but offset by decreases in outpatient spending.

Where do we go from here? With many virtual care flexibilities set to expire at the end of the year – like allowing Medicare patients to receive telehealth in their homes – regulators are on the clock to create more permanent policies.

  • Policymakers have already proposed a bevy of bills to extend the flexibilities, but the debate carries on as the deadline looms.

The Takeaway

Given the access benefits, quality improvements, and modest increase in spending, this study only makes it harder to justify rolling back telehealth coverage for Medicare patients. The evidence is mounting, and it’s not too hard to picture a world where the arguments against telehealth only grow weaker as technology improves and providers optimize their virtual care services.

Included Health Lunges Into Virtual Specialty Care

Where do you go when you’re already tackling primary care, behavioral health, and second opinions? If you’re Included Health, you swim straight upstream to specialty care.

The three bullet history lesson on Included looks something like this:

  • In 2021, Grand Rounds (virtual second opinions) and Doctor on Demand (on-demand visits for physical and mental health) merged to provide integrated virtual care.
  • The combined company entered care navigation, and now provides benefits guidance, financial support, and advocacy services to more members than possibly every other navigator in the country.
  • To make sure every patient felt this care was built just for them, the company acquired Included Health (culturally aware care for underserved communities), and here we are.

Specialty care is a natural continuation of that vision, starting with a trio of centers prepped for a 2025 debut: the Cancer Center, the Center for Women’s Health, and the Center for Metabolic Health. 

  • Members will have access to a specialist-led care team with integrated primary and mental healthcare, as well as in-home support for prescriptions, diagnostics, and monitoring. That all happens with a single member record, one central form of billing, a unified medical history, within the context of available benefits.
  • It’s a giant undertaking, but it’s made possible by roots in expert second opinions that helped grow a nationwide network of 4,000+ specialists and collaborations with over 40 health systems (not to mention a seriously well-rounded partner ecosystem) . 

The real strength of the expansion lies in the fact that it isn’t so much a new service line as a massive enhancement to a platform with all of the primary care, behavioral health, and navigation ingredients already baked in.

  • Outside of large orgs with the resources to try and coordinate the same breadth of offerings, few companies have had the scale, track record, and plain-old resources to cross the chasm into actual specialty care delivery. 

That means Included’s real competition will likely be Included. Can it scale without sacrificing quality? Can it deliver care for “all” that feels like care for “you”? Can it connect the dots between virtual specialists and the in-person care that will inevitably come into play?

The Takeaway

Included Health’s specialty care expansion hits the center of the bullseye with both patients seeking more cohesive care journeys and partners looking to right-size inflated portfolios of point solutions. That strategy also happens to be a good way to land on the shortlist of healthcare companies poised for a major IPO.

Scaling Adoption of Medical AI

Medical AI is on the brink of improving outcomes for countless patients, prompting a trio of all-star researchers to pen an NEJM AI article tackling what might be its biggest obstacle: real-world adoption.

What drives real-world adoption? Those who have been around the block as many times as Dr. Michael Abramoff, Dr. Tinglong Dai, and Dr. James Zou are all-too familiar with the answer… Reimbursement makes the world go ‘round.

To help medical AI developers get their tools in front of the patients who need them, the authors explore the pros and cons of current paths to reimbursement, while offering novel frameworks that could lead to better financial sustainability.

Traditional Fee-for-Service treats medical AI similarly to how new drugs or medical devices are reimbursed, and is a viable path for AI that can clear the hurdle of demonstrating improvements to clinical outcomes, health equity, clinician productivity, and cost-effectiveness (e.g. AI for diabetic eye exams).

  • Meeting these criteria is a prerequisite for adopting AI in healthcare, yet even among the 692 FDA-authorized AI systems, few have been able to pass the test. The approach carries substantial risk in terms of time and resources for AI developers.
  • Despite those limitations, FFS might be appropriate for AI because health systems are adept at assessing the financial impact of new technologies under it, and reimbursement through a CPT code provides hard-to-match financial sustainability.

Value-based care frameworks provide reimbursement on the basis of patient- or population-related metrics (MIPS, HEDIS, full capitation), and obtaining authorization for medical AI to “count” toward closing care gaps for MIPS and HEDIS has been shown to be considerably more straightforward than attaining a CPT code.

  • That said, if a given measure is not met (e.g. 80% of the population must receive an annual diabetic eye examination), the financial benefit of closing even three quarters of that care gap is typically zero, potentially disincentivizing AI adoption.

Given the limitations of existing pathways, the authors offer a potential new approach that’s derived from the Medicare Part B model, which reimburses drugs administered in an outpatient setting based on a “cost plus” markup.

  • Here, providers could acquire the rights to use AI, then get reimbursed based on the average cost of the service plus a specified margin, contingent upon CMS coverage of a particular CPT code.
  • This model essentially splits revenue between AI creators and users, and would alleviate some of the tensions of both FFS and VBC models.

The Takeaway

Without sustainable reimbursement, widespread medical AI adoption won’t be possible. Although the quest continues for a silver bullet (even the authors’ revenue-sharing model still carries the risk of overutilization and requires the creation of new CPT codes), exploring novel approaches is essential given the challenges of achieving reimbursement through existing FFS and VBC pathways.

Epic vs Particle: The Data Exchange Debate

It probably would have been impossible to wander onto any of your healthcare newsfeeds last week and miss the drama unfolding between Epic and Particle.

If for some reason the solar eclipse blacked out your internet, the basic timeline looks like this:

  • April 8 – Rumors began circulating that Epic cut off data access to patient information platform Particle Health.
  • April 9 – Particle confirmed that Epic ceased responding to medical record requests through the Carequality network (updates ongoing).
  • April 10-11 – All hell broke loose.
  • April 11 – Epic released an Issue Notification detailing the issues and steps toward a resolution.
  • April 12 – Our shepherd through the dark forest of interoperability, Brendan “Health API Guy” Keeler, published a masterful breakdown of the situation and its downstream implications.

Without wading too far into the data exchange weeds, Particle “combines data from 270 million plus patients’ medical records by aggregating and unifying healthcare records from thousands of sources”… sources like Carequality.

  • Carequality is effectively one of the country’s largest health information networks, facilitating data exchange between qualified network members (i.e. Particle) who agree to only query patient data for “Permitted Purposes” such as Treatment, Health Care Operations, or Public Health Activities.
  • The problem at the heart of the Particle controversy arises due to the fact that Treatment is the only purpose that organizations like Epic are actually required to respond to, causing all sorts of companies to warp their true purposes to Treatment-shaped requests.

Epic’s Issue Notification went as far as specifically naming certain Particle customers that it felt violated the Treatment case, including a company named Integritort that was allegedly using the patient data to try and identify potential class action lawsuit participants.

  • Particle maintains that all of its partners directly support Treatment, and that “the ability for one implementor to decide, without evidence or even so much as a warning, to disconnect providers at massive scale, jeopardizes clinical operations for hundreds of thousands of patients as well as the trust that is so critical to a trust-based exchange.”

The Takeaway

Since we know a perfect takeaway when we see one, we’ll leave it to the Health API Guy to wrap up the story:

“The tactical actions and who’s right or wrong really isn’t that important. Instead, they can serve as a catalyst and accelerant for the change needed. These events occurred because fraud and abuse are happening because the status quo of the networks only working for Treatment leads to the worst possible incentives. Health data is needed by a broader set of stakeholders in order to serve the patient.”

In other words, now’s the time to make viable paths for other Permitted Purposes a reality.

Rock Health Q1 2024 Funding Recap

If Rock Health’s Q1 2024 Digital Health Funding Report makes one thing clear, it’s that the times of transition are behind us, and we’re now fully entrenched in a new digital health funding cycle.

The first quarter saw $2.7B in digital health venture funding across 133 rounds, marking the lowest quarterly total seen since 2019. It’s not as bad as it sounds.

  • Although total funding wasn’t trending in the right direction, the pace of the funding was actually healthy, and the 133 rounds was the highest in the last six quarters.

The average transaction size of $20.3M mostly tells the tale of growth-stage companies having to justify their valuations based on clinical outcomes rather than fancy storytelling.

  • Crowded markets are pushing enterprise customers to seek out outcomes data as a way to differentiate players and evaluate value-for-investment.
  • As outcomes data becomes a moat and a customer draw-in, investors are seeking out companies that can demonstrate efficacy early – making outcomes data more central to fundraising conversations… and at earlier stages.

AI drove a record share of funding. While not exactly too surprising, AI-enabled companies captured 40% of Q1’s funding total ($1.1B across 45 deals), compared to 33% of 2023’s funding pot and 29% of 2022’s.

  • As AI energizes the sector, it isn’t too hard to follow the funding to the areas with the most perceived promise: scribing, precision medicine, and care enablement.
  • Abridge raised a colossal $150M Series C, AI precision health company Zephyr AI landed a $111M Series A, and a suite of high flying startups landed huge rounds, including Ambience Healthcare ($70M), Fabric ($60M), and Codametrix ($40M).

The last theme is familiar: creative fundraising continues to be a crowd favorite, especially as public market delistings cause investors to rethink their exit potential.

  • Nearly half (48%) of Q1’s funding rounds were unlabeled, compared to 44% of all transactions in 2023.
  • Founders are going above and beyond to entice investors with more upside in the event of an exit, as seen with Transcarent structuring its $125M Series D to offer funders 2.5x their investment should the company M&A or IPO. 

The Takeaway

Expectations have been reset for digital health startups, causing them to evolve their strategies and reorient around different metrics of success (strong outcomes / margins vs. high projected growth). These expectations are undoubtedly higher than they were during the pandemic era of loose capital, but that’s probably not a bad thing for a sector that’s still striving for maturity.

Providence Spins Out Praia Health After Series A

The rockstar team over at Providence’s Digital Innovation Group officially spun out its fourth digital health startup: Praia Health.

Praia Health is making its grand debut armed with $20M in Series A funding to help hospitals avoid the commoditized care caravan by building deeper relationships with their patients.

To accomplish that, Praia enables the creation of robust consumer profiles that extend beyond the medical record, which might finally give providers a way to build the “digital flywheels” enjoyed in other consumer industries.

Here’s how it works (perfectly explained in these short videos):

  • The first step is a “lift and shift” to Praia’s Secure Patient Identity Service (SPI). Once transitioned, SPI seamlessly supports all of the system’s signed-in digital experiences (branded mobile and web apps, patient portals, etc.).
  • Praia Health’s PersonStore then enables health system’s to marry consumer identity with consumer data, securely synchronizing the EMR with outside data sources to connect-the-dots between consumer data and outcomes. 
  • Those SPI and PersonStore capabilities lay the groundwork for a new class of digital experiences that reflect the health system’s unique brand and offerings, assisted by a full suite of development tools to integrate Praia into existing solutions.
  • Once those consumer experiences are available, the flywheel is fueled by adding more solutions to Praia’s partner ecosystem. The platform delivers all of these solutions through a single scalable experience.

Although Praia has some lofty goals, Providence’s incubation model carries some massive advantages that might make them possible.

  • Praia has been in use at the 51 hospital system since January 2022, and already supports over 3.5 million user accounts.
  • Indiana-based Community Health Network is already lined up as Praia’s first post-launch implementation.
  • The aforementioned partner ecosystem is debuting with a whopping 17 startups on-board, including Omada, Wellthy, Kyruus, and Season.

The Takeaway

Although there’s no doubt that Praia’s mission is ambitious, it’s equally apparent that Providence brings plenty of advantages that other founders can only dream of. Providence has a solid track record with its incubator (see: Xealth, DexCare, Circle), and all signs are pointing to Praia keeping that streak alive.

The Cost of Manual Workflows, and Where Automation Can Help

We cover new solutions promising to swap manual workflows for automated operations on a weekly basis, but it’s rare that we get a chance to devote a full deep dive to the motivations driving the innovation:

  • What workflows would see the biggest benefit from automation?
  • What areas are healthcare orgs hoping to evolve in their operations?
  • What teams stand to gain the most from automation? 

Medallion’s 2024 State of Payer Enrollment and Credentialing Report gave us that chance, shedding light on those answers through a survey of nearly 350 healthcare executives.

Among the key findings from the report was the fact that 45% of respondents say their staffing levels are “inappropriately low,” yet 34% also feel the need to further cut headcount expenses.

  • Those numbers are a recipe for burnout. 57% of enrollment and credentialing teams have experienced turnover in the past year, along with 36% of CNAs, 15% of NPs, and 11% of PAs.

One of the main reasons why enrollment and credentialing teams are feeling the pressure so acutely is because of the manual nature of their workflows.

  • The payor enrollment process involves wrangling information from providers to fill out applications, staying on top of the evolving requirements of various health plans, communicating the enrollment status of every provider, and constantly following up with payors by email or fax. 
  • Those slow turnaround times directly impact the bottom line, with 46% of respondents reporting that unoptimized enrollment workflows cause them to miss out on revenue.

That isn’t even half the battle, with the credentialing process taking just as long to gather provider data, check qualifications, and complete primary source verifications.

  • 84% of credentialing teams experience turnaround times of 15 days or more, which unfortunately isn’t too surprising considering that 30% manually verify credentials by visiting individual sites.
  • Every day wasted waiting on credentialing is a day the provider isn’t seeing patients, and that missed reimbursement turns out to cost an average of $10k each day.

If those problems hit a little too close to home, automation is more than likely going to play a major role in the solution. An end-to-end platform like Medallion might be the right way to make that happen, and you can check out our coverage of the platform for a complete overview of how it can help fully automate payor contracting and enrollment, credentialing, and licensing.

The Takeaway

It’s no easy task to balance operational costs with a mission to provide high-quality care, but with the US healthcare industry spending over $800B every year on administrative tasks, it’s time to find a way to thread that needle. For a closer look at these issues and how Medallion might be able to help, make sure to head over to the full report.

Get the top digital health stories right in your inbox

You might also like..

Select All

You're signed up!

It's great to have you as a reader. Check your inbox for a welcome email.

-- The Digital Health Wire team

You're all set!