...
By Alexis Kayser
Healthcare Editor
0
Share
Newsweek is a Trust Project member
See more of our trusted coverage when you search.Prefer Newsweek on Googleto see more of our trusted coverage when you search.
Hospital for Special Surgery (HSS) has a catchphrase for its AI strategy: "Stay in the game, win the game, change the game."
Ambient scribes are no longer changing the game, Dr. Ashis Barad, chief digital and transformation officer for the New York City–based system told Newsweek on July 18. They have become a necessity to "stay in the game" and are an expectation of the clinical community.
To win health care's AI game, health systems need to use technology to automate and augment more than just the patient visit. As the race for patient preference picks up, those that can completely revolutionize the care experience are going to come out ahead.
Read More
Calling All Health Care Leaders: Join a Discussion on Health Tech’s Future3 min read
When City of Hope Launched Its Own AI Tool, Execs Got Hugs and Pharma Came Calling5 min read
What AI Really Can Do Now: 6 Lessons for Harnessing Artificial Intelligence18 min read
"What we're doing in the core business today needs to be fundamentally different going forward," Barad said.
It's going to be a hefty undertaking. As vendors and health system leaders look beyond the quick wins of earlier iterations of generative AI, they expect to face a number of regulatory, infrastructure and reimbursement challenges, sources told Newsweek.
...
What does the next phase of health care AI look like?
Historically, health care leaders have had to make forward-looking decisions based on retroactive data. AI can be used for forecasting and predictive modeling that will allow health systems to draw upon a "central brain" and adjust their strategy in real time, Barad said.
Some health systems are already using predictive modeling to coordinate staff, reduce admissions delays and deploy equipment. Approximately 300 hospitals across the globe use GE HealthCare's Command Center technology to improve operational efficiency.
"We're not relying on word of mouth or pagers or systems that don't work well," Kristie Barazsu, president and COO of Duke Health Lake Norman Hospital in Mooresville, North Carolina, told Newsweek of the system's Command Center experience. "There's one source of truth."
Innovaccer—an AI platform used by more than 100 U.S. health systems, including seven of the 10 highest-revenue health systems—recently launched its Readmission Management tool, using predictive modeling to identify expensive, avoidable patient readmissions. The solution includes AI agents that can proactively reach out to patients who are at a higher risk of readmission, preventing them from falling out of touch with their health care teams.
Abhinav Shashank, co-founder and CEO of Innovaccer, told Newsweek that the company is deliberately shying away from "point solutions" (individual tools that tackle individual issues) in favor of maximizing capabilities on a single platform.
Moving forward, "everyone will want a platform that gets these AI capabilities to stick together and happen in a coordinated way, rather than in this haphazard way that is going to happen if we let point solutions proliferate," he said.
Indeed, continuity is a big part of health systems' next play. Barad and his team are in the midst of an 18-month centralization effort at HSS, crafting a "lake house" architecture to house data in a single location.
This will allow the system to decouple data entry from data storage without disrupting employee workflows, Barad said. Through HSS' partnership with Amazon Web Services and Databricks, it is working to funnel data from any initial entry point—whether that's Workday, Epic or another data system—into a central pool where it can provide a more comprehensive view of system operations and patient journeys.
Once all of that data is in one place, it will be easier to deploy AI-based analytics to tackle pain points at the enterprise level. It's a way to break down siloes across the brand without asking employees to change their behavior or introducing new data entry systems. For health systems like HSS, that seamless data flow is essential to creating a seamless patient experience.
Ultimately, HSS wants to be like Disney, according to Barad. They are currently known for their world-class hospitals on Manhattan's Upper East Side, providing episodic surgical care. Those hospitals are like a medical version of Walt Disney World, drawing international visitors and providing an all-encompassing experience while you're within their walls.
But Disney remains part of everyday life, even after you leave the theme park**.** The brand is a constant part of the public consciousness, through movies, streaming services, music and merchandise.
"We're so amazing in our Disney World, but we've stayed in Disney World," Barad said. "How do we broaden and be part of your high school physical or yearly checkup?"
He sees AI as an answer to that question—a way to improve the connection with patients and be present in their health journeys before pain appears—without deploying limited human resources. As wearable devices and fitness trackers become more mainstream, health systems could soon have access to more data than ever before. AI agents can be particularly helpful in making sense of that constant information flow, using it to prevent patients from falling through the cracks between episodes of care.
HSS is working with Palantir to build a full orchestration layer that will bring all of this data together, giving the health system a horizontal view of the patient experience: from the moment they say "ouch" to the moment they schedule an appointment to the moment their insurance authorizes payment.
"Up to this point, nothing has orchestrated all of those assets that we have in the ecosystem into a continuous journey for a consumer to go through," Barad said. But doing so "actually allows for there to not be any handoffs or drop-offs, and the outcomes are better because of it."
The health system has also hired new talent to tackle the complex challenges ahead, including Chief Data and Technology Officer Dr. Michael Apkon.
"This new world has to be built," Barad said, "and it has to be built by people who know what they're doing."
What challenges does health care still face when implementing AI?
The excitement of ambient scribes and AI chatbots has started to wear off, and health care is running out of quick wins as it looks to construct this "new world."
While AI promises efficiency, health systems will have to put in significant time to unlock its full benefits.
Beyond note-taking tools, there are "very little success stories" of clinically implemented AI, said Dr. Regina Barzilay, distinguished professor of AI & Health in the Department of Computer Science at MIT's School of Electrical Engineering and Computer Science and the AI Faculty Lead at MIT's Jameel Clinic.
She told Newsweek that there are abundant stories of new AI algorithms that show promising results but few that have actually proven themselves beyond mere promise.
"You need to know how to safely bring it into the [health care] system," Barzilay said. "There is a new science, which is the science of implementation."
As with all sciences, implementation is complicated. Each hospital has a different experience when translating AI into practice, due to dramatic variations in their infrastructures. The tools are still new but touch multiple departments in a health system, which makes it hard to find a single point person who can lead the strategy.
Plus, AI is rarely mentioned in clinical guidelines that govern physicians' decisions, according to Barzilay. For example, the National Comprehensive Cancer Network (NCCN) keeps practice guidelines for 72 types of cancer. Only one—the guideline for prostate cancer—makes any mention of AI.
"The clinician who treats you has to follow what is considered to be standard according to the guidelines," Barzilay said. "AI-powered solutions are not part of the guidelines, so clinicians will not use it."
Then, there's the question of reimbursement. If an AI tool recommends that a patient needs a follow-up appointment, will insurance companies consider it sufficient evidence to pay for that visit? Few AI tools have managed to get their own diagnostic codes, making the reimbursement situation "a bit of a Wild West," per Barzilay.
Moving forward, she believes that clinical trials will be key to solving guideline and reimbursement issues. AI must be tested and validated in a controlled environment to prove its efficacy—just like prescription medications, novel therapies and surgical devices.
However, science moves slowly, and cuts to federal health agencies' funding are likely to hold up the process even further. "These studies are not cheap," Barzilay said, "so it's important to find a way to pay for them."
As health care looks to the next phase of AI, it's going to have to reimagine old systems, rewrite past guidelines and conduct long-term research. This is all going to take time.
And, unfortunately, the clock is ticking. The U.S. is already falling behind other countries when it comes to health care AI, according to Barzilay. It is easier to roll out new technologies across government-led or nationalized health care systems, like those in the United Kingdom, Israel and the Netherlands. Under these models, there is typically more uniformity among hospitals in the same country.
In the U.S., however, hospitals can have vastly different financial resources and digital infrastructures, even within the same state. Each health system has to more or less start from scratch when selecting and implementing its AI tools.
"I think the federal government really has to lead in this transformation," she said, "because it involves so many parties, and it costs so much money to complete the studies and understand how each tool actually changes the outcomes."
Barad hopes to eventually use AI to improve health equity—scaling the expertise of HSS' world-class surgeons to even the smallest rural hospitals. At this point, codifying an AI agent with eons' worth of medical knowledge is the "easy" part. To stretch the result across state lines and siloed systems, hospitals are going to have to change the game.
Want to hear more? Dr. Ashis Barad will be speaking at Newsweek*'s Digital Health Care Forum on September 16 in New York City, along with leaders from Kaiser Permanente, City of Hope, Sanford Health, Northwestern Medicine and more. Click here to learn more and secure your ticket.*
View 0 Comments
Submit Correction
View Editorial & AI Guidelines