Digital divide widens in wake of AI, machine learning

It has been more than a decade since President George W. Bush set out to get electronic health records for every American. In the 15 years since his pronouncement, there’s been significant implementation of EHRs across the country, propagating an incomprehensible amount of data.

It has been more than a decade since President George W. Bush set out to get electronic health records for every American. In the 15 years since his pronouncement, there’s been significant implementation of EHRs across the country, propagating an incomprehensible amount of data.

In many cases, that data sits dormant and untapped of its potential.

Some healthcare organizations contend that it’s financial constraints that create limitations. Lack of dollars makes it difficult for all but the most advanced and lucrative healthcare organizations to put machine learning or artificial intelligence in place to make the most of the data.

Other experts say the bottom line isn’t the only obstacle. “It’s not necessarily only a divide between those who have the resources and those who don’t,” says Michael Millenson, president of Health Quality Advisor. “There’s also a cultural divide between those who are willing to embrace the potential of information age medicine and those who have been dragged kicking and screaming into the electronic health record age but still, in their heart of hearts, feel about EHRs the way buggy whip makers did about the early automobiles— ‘Get a horse.’ ”

“Practically every provider has EHRs now, but many don’t use those data for quality improvement or for research and other things that could help them,” says Lucila Ohno-Machado, MD, professor of medicine and chair at the UC San Diego Health, Department of Biomedical Informatics. Not every medical center or practice has the ability to meaningfully analyze the data, even though some have built-in tools that can subset data in certain ways—organizations still need brain power, statisticians, data scientists and other capabilities. Applying more intelligence to data is an important asset of any institution, she says.

Machine learning is important to gain enough value for data analysis, experts say. “It’s huge,” says Dawn Paulson, director of health information management and practice excellence at the American Health Information Management Association. “It offers so many more opportunities. Everybody’s got data, and not everybody’s using it for what it can be used for,” she says, and that’s a shame, because “for so many years, we focused on capturing the data and capturing it the right way so we can produce something out of it.”

Paulson has observed that a lot of the larger healthcare organizations are dabbling in AI, but many of the smaller organizations “just don’t have the resources.” She is also seeing large organizations partner with companies to get things moving. A few examples include Memorial Sloan Kettering partnering with IBM Watson, to use machine learning on treatment plans for cancer patients; the University of Pittsburgh Medical Center teaming with Microsoft to improve clinician empowerment and productivity; and Rush University in Chicago working with Cloudera and MetiStream to use machine learning to customize treatment plans.

Challenges for small organizations
Randy McCleese says the mandates put in place over the past 10 years to move healthcare as a whole to electronic health records forced smaller hospitals to go digital because the vast majority of their income is from Medicaid and Medicare reimbursement, and adoption was linked to that reimbursement.

But that doesn’t mean they could afford to do so. Profit margins are razor-thin in rural areas, says McCleese, CIO of Methodist Hospital in Henderson, Ky., a 192-bed acute care hospital, 25-bed critical access hospital and 19-practice physician network. So for many smaller facilities, adopting EHRs, meant going without updating radiology or lab equipment, for example.

Smaller rural hospitals face the very real survival option of being absorbed by a larger organization and losing the hometown touch that patients want, McCleese says.

Other troubles also prohibit the use of advanced technologies such as AI. Many healthcare organizations have a tremendous amount of data in multiple legacy systems—all of which must be retained to ensure continuity of care. The storage of data on legacy systems is causing providers to be skeptical at what they are seeing, he says.

In addition, hardware on the legacy systems is old and difficult to maintain. And when maintenance is no longer feasible, data must be extracted from the old legacy systems and stored somewhere. These are common decisions for most organizations these days, McCleese says.

Making matters worse is the constant threat of breaches and ransomware attacks. “The whole security thing is exploding,” McCleese says, making cybersecurity another priority competing for limited resources.

Staffing is another huge issue, McCleese says. It’s difficult to lure top IT talent—or any IT talent that’s in limited supply—to smaller markets, both because highly skilled IT staff don’t want to live in the small towns, and the healthcare organizations in these areas can’t offer the salaries that matches the going rate in metropolitan areas.

It’s critical to get board level understanding of the importance of hiring top talent, he says. Sometimes organizations can develop their own talent, but that takes time. “CIO churn,” which is common in rural hospitals results in a stop-and-start pattern to growth, as each new CIO tries to leave his or her individual mark on the institution, often undoing some of the work previously done.

It’s critical that top leadership understand the impact that big data can have on the future of healthcare, and that’s not always an easy sell, McCleese says.

The bottom line is that money is only going to solve some of the challenges that smaller organizations face, according to McCleese.

“From my perspective, in smaller markets, we’re not seeing so much an occasion to use AI, as we are just trying to survive,” he says. It may be that the best option for a smaller organization to use AI is if they team up with a larger organization.

Health systems and AI
Omer Awan, CIO of Macon, Ga.-based Navicent Health, agrees that some organizations don’t have the money or other resources to push the boundaries on disruptive aspects of new-age healthcare that include AI.

However, organizations that fail to do so will regret it, he says, particularly because it won’t be long—less than a decade—until millennials will start flooding the healthcare system. “The way we interact with our consumers today is not what appeals to these people,” he says. For them, convenience takes precedence over everything.

So many decisions today can or should be based on data—that’s where the promise is, according to Awan. He gives the example of a large trauma center that uses an algorithm to determine which patients should be seen first by the physicians in the ER. That extra assistance not only frees up the physician to fully focus on the patient at hand, but it can often help to achieve the best outcomes.

Awan says money isn’t the only thing holding organizations back from making the most of AI. It’s also priorities—the internal mission.

Then, there’s the time factor. CIOs typically are consumed with the daily operations of a healthcare organization. The traditional work is overwhelming; it’s hard to add on one more thing, he says. “The digital stuff is very exciting, but sometimes an organization just doesn’t have all the resources to do it justice or to bring focus.”

Navicent recently gave Awan a new title, chief information and digital officer—with a budget soon to follow—and the goal to push the health system forward into more powerfully tapping data.

Awan says CFOs need to see proof of a return on an IT investment, if you want buy-in from the top. How will any new digital strategy bring ROI? Part of the problem is, CIOs can see how things might be improved with a new digital strategy, but they lack the ability or data to show a financial benefit on paper. The IT guy can see what’s happening and can promote the changes, but he has to think outside the box and sell it to the C-suite with actual financial projections, Awan says.

Vendor roles
There’s a “crazy disconnect of the untapped intelligence,” says Jeff Fritz, CEO of Revel Health in Minneapolis, a communications platform that offers machine learning services to improve outcomes, star rating measures for payers and health risk assessments.

“Healthcare generally is a decade behind the rest of the world,” Fritz says. “Because of that, there are these older structures in place, wrapped with high fear of security risk. They don’t tend to attract the teams of people that are the leading minds in technology, and so a provider or payer organization is at a disadvantage with modern thinking.”

Fritz doesn’t see a silver bullet coming along any time soon. Right now, in healthcare, innovators are still trying to knit together old and new technology—which optimistically, should lead one day to innovation. “Solutions outside of old technology will come up with progressive answers. We will have to work on the solutions,” he says.

According to Fritz, the digital disparities among his clients are obvious. Larger payers are all in acquisition mode. “The big are getting bigger,” and this is leading to more sophisticated technological abilities. He likens these organizations to swimmers who are in the deep end of the pool. But for those organizations in the shallow end, it’s too difficult to find an easy remedy. That’s where Revel Health, like other vendors, help bridge the gap. “The smaller organizations see the results we can provide. Smaller payers don’t really have a deeper end of the pool,” he says. “They don’t have the sophistication to leverage AI.”

Nonetheless, Fritz says most of Revel’s clients are very large payers, spanning the U.S., in addition to some regional payers, ACOs and managed care organizations. Revel uses AI to make communications more personal, using social determinants and other factors.

For example, Revel Health recently worked with 750,000 patients to ensure they received flu shots. But instead of contacting them all the same exact way, it used AI to develop up to 100,000 ways, depending on the various data about the patient. The individualized contact plans resulted in a 30 percent increase in flu shots for Medicare patients and a 50 percent increase for those insured by commercial payers.

More for you

Loading data for hdm_tax_topic #better-outcomes...