AI and other tech show promise in democratizing healthcare
The use of advanced computing and virtual care technology has demonstrated value, and wider application can improve care, HIMSS panel contends.
There’s growing hope that artificial intelligence can help enable healthcare providers in the rising call to give historically underserved populations a more equitable share of technologically supported care.
Continued use of technologies that received wider use during the pandemic, particularly telehealth, also can provide more access to level the playing field for care. But AI shows great promise in being able to democratize care, supporting population health and easing burdens on clinical staff, according to presenters at a panel discussion on the topic of “Democratizing Healthcare with AI” Tuesday morning at the HIMSS Annual Conference and Exposition in Orlando.
The discussion was sponsored by H2O.ai, a startup that offers an open-source framework and proprietary apps that make it easier to build and operate artificial intelligence-based services.
Broader application, more equitable care
Artificial intelligence got off to a rocky start several years ago, with potential benefits being overhyped and with some preliminary reporting suggesting that it could usurp clinical and diagnostic roles. That’s died down, and now it’s proving some worth in a variety of supportive roles that make sense for computer-assisted support.
“Building an AI pipeline builds better access for patients."
For example, in one use case, says Bob Rogers, CEO at Orchestrated Intelligence and expert in residence for AI at UCSF Smarter Health, a center for digital health innovation at UCSF. For example, UCSF partnered with H2O.ai on using artificial intelligence to interpret the 1.4 million faxes it received each year, interpreting the text in the faxes to better funnel and accelerate processing, especially for communications regarding patient referrals.
“Our default has always been to go to a human process,” Rogers said, referring back to how faxes were traditionally handled as just one example. “Building an AI pipeline builds better access for patients. It’s really opening doors for how we can make all of our processes more efficient.”
Medicare & Medicaid, reducing costs
AI can also make a case in the area of cost control, which will be important in gaining wider support and federal funding, said Joel White, president and CEO of Horizon Government Affairs, a consultancy specializing in actionable strategic and tactical advice for navigating the congressional and regulatory processes.
“There’s a lot of interest in Congress on how to get costs under control. They understand AI could make an impact; the question I get all the time from members of Congress is how,” White said. The proven use of AI in screening for early signs of diabetic retinopathy, with documented savings from early intervention, is something that resonates, he added. “I can take that back to Congress all day.”
“Data is kind of messy in Medicare and really messy in Medicaid”
One place to make the case for using AI is in the federal government itself, specifically in the Department of Health & Human Services, which sees more than $1 trillion in health payments flowing through it. It expends some resources on finding fraudulent claims, but wider use of AI could help identify these as they come in, even with data challenges posed by Medicare claims, White said.
“Data is kind of messy in Medicare and really messy in Medicaid,” he said, referring to the myriad forms of claims received by HHS. AI started to be used to identify fraud in telehealth claims during the pandemic, and there’s no reason not to use it more broadly, he added. “The tools are available – AI works really well on messy data.”
Making a beachhead with demonstrable improvements or savings can help expand the vision for funding AI in other ways that can improve healthcare more broadly across a wider population.
Using advanced computing technology has untapped potential to take costs out of the system, which can result in those funds being reallocated to actual patient care, suggested Fred Goldstein, president and founder of Accountable Health, a consultancy focused on population health, health system redesign and analytics.
Particularly in population health, AI and analytics can help achieve better segmentation of patient populations and, when matched against other patient-specific capabilities, such as genomics, can help to not only find individuals with conditions but also to focus in on the most effective treatment option. “Then you can get to an N of 1, and know exactly what I need to do. We can use AI to help focus in on one person at a time.”
AI has already helped in some limited cases to get care to more patients. For example, it was used in India to optimize vaccine distribution across its vast population. Even so, AI is not a magic bullet in democratizing care, and it’s important to identify bias in the information the technology uses to reach its conclusions, Goldstein acknowledged. “It’s important to focus on health equity, so what we are building – ensuring that what comes out of the black box is correct – so that we’re not influencing healthcare in a negative sense.”
“What we’re really talking about is explainable AI,” Rogers added. “We’re really talking about trust, the thought process that went into it, the pieces of information that went into it. Trust is more than being able to explain – it goes to what is the question we are trying to answer.”
“It’s a tool,” White concluded. “When it’s used to demonstrably improve people’s lives at the end of the day, it’s improving healthcare.”