PARTNERED WITH  |
|
|
Happy Friday Jr., Hospitalogists!
This week we're digging into two AI headlines, a good read on denials, and a recap of takeaways from the MA ways and means hearing.
Let me know what you guys think. Let's get into it! |
Was this email forwarded to you? |
|
|
SPONSORED BY TEAM CONNECT Starting January 1, 748 hospitals will enter TEAM, the 5-year, mandatory bundled payment model for 5 surgical episodes. Many of the hospitals chosen by CMS have no prior experience with bundled payments. TEAM Connect is here to help! TEAM Connect is a collaborative community created to help TEAM-participant hospitals navigate episode-based payment models. Membership, which is free for hospital leaders, ACOs, and physicians, provides access to webinars, discussion forums, and a national network of healthcare professionals who are operationalizing bundled payment models for their hospitals. Corporate members can also join for a membership fee and provide their expertise in value-based care. TEAM Connect's September and October webinars are just around the corner, so don't miss out! Create an account and join today! |
|
|
ICYMI: HEALTHCARE EDITION |
Healthcare headlines, trends, and good reads you may have missed. |
The AI Scribe Wars were Never about Scribing |
Just one week before Epic announced their AI Scribe (powered by Microsoft Dragon technology, mind you), Abridge announced a partnership with Highmark on real-time prior authorization. Timing is everything. Epic's AI Scribe launch has people declaring the death of ambient AI valuations. As always, this is a gross overreaction. But they're half right: 90% of AI Scribes are toast because nobody needs 130+ companies doing the exact same thing. The smart money isn't panicking. AI scribe valuations were never about transcription quality. No, as we've mentioned, the future state of ambient sits with workflow infiltration. Take a peek at recent large, well-funded ambient announcements: - Ambience characterized itself as an 'end-to-end platform for documentation, coding, and clinical workflow support that operates across the full continuum of care.'
- Abridge, noting the above, launched its Contextual Reasoning Engine and specifically called out shifting upstream into risk adjustment along with coding capabilities in its Series E announcement
- Nabla partnered with Navina - a dynamic duo and is expanding into agentic AI for complex workflows, beyond documentation: 'Now, Nabla is expanding beyond documentation into a more agentic model of clinical AI. This next phase enhances clinical documentation integrity (CDI), initiates EHR actions, and adapts across care settings to support diverse clinical roles. By unifying ambient listening, dictation, coding, and command capabilities into a single extensible agentic platform, Nabla is building toward its long-term vision: a proactive assistant that intuitively streamlines existing workflows.'
- Finally Commure launched Commure Agents over the summer and also had plenty to say around AI-enabled RCM capabilities after its $200M raise. Don't forget Commure is partnered with two of the largest, most profitable health systems in Tenet and HCA.
Once you become trusted software in a physician workflow, you can go upstream and downstream into literally any other workflow. Plus…who's going to be the one to tell the physician they have to switch their documentation software? I don't know about you but I'm a Google Maps truther. I'm not about to switch to Apple. Retention and stickiness of existing physician and clinician users has to be incredibly high. Abridge doubling down on revenue cycle management (RCM) workflows with Highmark is a perfect example. Their AI solution automates completing prior auth forms, submitting and reviewing requests, tracking status, and identifying when authorization is needed in the first place. And Highmark is an interesting enterprise, operating both a payor arm along with a large health system. It's a great setup to test AI deployments around payor-provider convergence. Ambience's leaked deck shows the same strategy: their $243M Series C is betting on workflow expansion, not better dictation software. |
So no, the AI scribe valuations won't go to zero. They'll consolidate around companies that find ways to go upstream or downstream where actual money is made (or saved). But, we should note an important Nuance (only the OGs will understand this reference): going upstream requires deep EHR partnerships AND finding use cases EHRs don't want to build themselves. Prior auth is perfect: too messy and politically sensitive for Epic to own directly. But the point stands: EHR product roadmaps are always a looming threat. In healthcare, distribution trumps innovation. Abridge figured this out early, went all-in on Epic partnership (they literally opened a Wisconsin office to be closer to headquarters and from what I - Blake - understand, Epic has equity in Abridge as part of their partnership), and now they're reaping the benefits. TL;DR: Epic will absolutely massacre most AI scribes. In my (Blake) opinion, it's a play to capture the long-tail of health systems on Epic that haven't necessarily rolled out a deployed scribe solution, while the Abridge's of the world (also partially owned by Epic) spread their tendrils throughout health systems as AI vendors of choice for exponentially increasing use-cases. And this give by Microsoft is perhaps a signal of a couple things: Nuance has been losing ground in the scribe wars while Microsoft and Epic more closely align between cloud and AI capabilities. Either way, Epic wins. And they always will. But just keep in mind: this announcement was never about note-taking in the first place. If you're still doing only that as an AI scribe, you've lost. |
OpenEvidence Aces the Test. Is it Enough? |
OpenEvidence scored 100% on the USMLE. But that doesn't actually matter that much. OpenEvidence, basically UpToDate meets ChatGPT that 40% of US physicians use for clinical decision support, just announced they achieved a perfect score on the United States Medical Licensing Examination. ChatGPT-5 "only" scored 97%. First of all: interesting to think that OpenEvidence's $3.5B valuation is built on scoring just 3% better than OpenAI's general model, while OpenAI went from 60% → 97% accuracy in three years. Second of all: while a 100% USMLE score is impressive, it's actually not that meaningful for real clinical practice. A study just came out that tested AI models on completely fictional medical content - they created an imaginary organ called the "Glianorex" and generated medical textbooks and exam questions about it. AI models averaged 64% accuracy on these fictional questions, while actual physicians scored only 27%. (both needed to guess the right answer as they had zero knowledge about this fictional organ) When researchers analyzed how AI was getting answers right, they found it was using: - Test-taking strategies: Picking specific answers over general ones, avoiding absolutes like "always/never"
- Pattern matching: Applying real medical principles to fictional conditions
- Hallucinations: Making up plausible-sounding medical facts
The study concluded that AI excels at pattern recognition and exam heuristics, not genuine medical reasoning. This is why we need better evaluation methods - complex clinical vignettes that require actual diagnostic reasoning, not multiple-choice pattern matching. Because real medicine is messy, and patients don't come with multiple choice options. PS: OpenEvidence's founder, Daniel Nadler (PhD) agrees that a USMLE score is not the best way to evaluate OpenEvidence in a clinical context. |
What the Medicare Advantage Ways & Means Hearing actually Means |
A couple of weeks ago, the joint Subcommittees on Health and Oversight evaluated the last two decades of the Medicare Advantage program. TLDR; The political center is shifting. Key Takeaways: - The MA program continues to grow (54% of beneficiaries in 2024)
- MA costs 20% more per enrollee than traditional Medicare and critics argue value (like more coordinated care, supplemental benefits, and chronic disease support) doesn't match cost
- Even traditional GOP supporters are calling for reforms to curb overpayment and abuse by large MA insurers
- Prior authorization overhaul is needed as high denial rates, delays, and inconsistent plan rules are harming patients and providers
- Risk adjustment and star rating reform is needed to ensure payment reflects measurable outcomes
- Transparency needs to be a cornerstone, particularly around AI usage, denial decisions, and financial incentives
- Standardized payments are key to retaining providers
- Rural network gaps threaten care access where it matters most
What's next? Several proposals to increase MA transparency are already gaining traction: - Enhanced MA oversight and audits, targeting risk adjustment abuse and denial practices
- Prompt payment and network adequacy standards, especially for post‑acute and rural care
- AI usage reporting and standardized prior authorization systems
It's definitely time to start preparing for changes on billing practices, provider networks, and even how AI is regulated in healthcare access decisions. We could see concrete moves to overhaul prior authorizations, inject transparency into plan operations, and realign MA incentives as soon at the fall session. |
The Denials Villain Outside Hospitals' Contracts |
I've been talking about payor price transparency a ton lately, and this read from Kaufman Hall caught my eye. It's about the payor provider manual, AKA, something we rarely talk about. Big Picture: It's not just about getting paid, it's about making the margin work. Denials are draining hospitals. Not just in dollars, but in strategic bandwidth. Traditional revenue cycle tactics like better documentation and appeals management are helpful, but they're ultimately just playing whack-a-mole with symptoms. KaufmanHall argues that the real issue is weak contracting and a lack of alignment between payer expectations and provider operations. The payor's provider manual lives outside the contract, is updated frequently (often without notice), and dictates how hospitals get reimbursed. Many denials stem from noncompliance with these constantly shifting manuals—rules that hospital teams often don't know have changed. Key takeaway: Read the fine print. - Inventory contracts and provider manuals. It's a pain, but tracking manual updates across all major payers helps mitigate risks hiding outside of contract terms.
- Negotiate contract language around manuals. Specifically, language around provider manual updates, notification timelines and payer accountability.
- Make sure managed care and rev cycle talk to each other. Make sure they share data, too.
|
|
|
SPONSORED BY CEDAR The One Big Beautiful Bill presents an ugly financial crisis for providers. But Cedar is here to help! Cedar developed a robust policy brief outlining: How these changes will directly impact hospitals and health systems The technology that can create a "digital safety net" to protect your margins and your patients What you can do to prevent overwhelming already-strained teams A step-by-step framework to identify your biggest risks now Don't wait to start capturing revenue and cushioning against looming payment cuts and economic pressures. |
|
|
Thanks for the read! Let me know what you thought by replying back to this email. — Blake |
|
| .png) | Share Hospitalogy, Earn Rewards | Have friends who'd love Hospitalogy too? Click the link below to share Hospitalogy with your friends and earn awesome rewards! | |
|
PS: You have referred 0 people so far | | Share Hospitalogy! | |
|
|
|
|
Get your brand in front of 48,000+ executives and healthcare decision-makers. |
I'm building a community of leaders in strategy, finance, and ops
at hospitals and health systems to help us connect, learn, and grow together. |
Workweek Media Inc.
1023 Springdale Road, STE 9E
Austin, TX 78721 Want to ruin my day? Unsubscribe. |
|
|
|
No comments