AI and EHR: Perfect Together?
Big data and electronic health records — along with other technologies — could change the way long-term care is delivered.
We’re living in an age of unprecedented technology. Back in the 1960s, when most of the residents at your long-term care facility were establishing their careers and rearing children, the first computers began appearing in workplaces around the country. They took up entire rooms and crunched numbers via actual paper ballots. Today’s children wouldn’t recognize them as computers. But steadily, over the next 60 years, these machines shrank in size and grew in their ability to not only dominate our lives, but potentially help us live longer, better, healthier lives, too.
Now, most of us have a small computer in the shape of a smartphone sitting in our pocket all day long every day. The computing power in these tiny devices would have taken a staggering amount of floor space just a few decades ago. Today, we can access a wealth of information on any conceivable topic in a split second. What if that computing power could be harnessed for good, to improve the health of older Americans residing in senior care facilities around the country?
Many facilities are doing just that, and often this push to improve patient care through intelligent analysis of data starts with the transition to electronic health records (EHRs). Yes, much has been written about how challenging the conversion to e-records has been for many health care intuitions across the spectrum of care. But with each facility that comes online, our collective mountain of data grows, and within that data may lie the future of clinical care.
Dr. Daniel Walker, PhD, MPH, assistant Professor of Family Medicine in the Institute for Behavioral Medicine Research at The Ohio State University Wexner Medical Center says that when we talk about electronic health records, “we’re talking about systems that are implemented into the clinical care process for both clinical and billing processes. They record all the information about the patient and their clinical condition and they support the coding processes on the back end,” so that facilities and providers can be paid for their important work.
Brian Geyser, APRN-BC, MSN, chief clinical officer, at Inspīr, a senior living facility in Manhattan adds that EHRs “are simply a way to digitize huge amounts of information and organize it in a way that provides actionable, clinical insights. From those insights, we’re able to develop strategies and targeted care plans. In the past, we had to write it down and didn’t have easy access to the data. But now, EHRs are becoming more sophisticated and intelligent,” as companies are beginning to conduct more analyses.
Geyser says AI and computer algorithms are just at the beginning of how they’re going to revolutionize the way clinical care is delivered in long-term care facilities, and the way care is delivered may look quite different in another decade once these systems become more developed.
Still, making the switch hasn’t been easy for everyone, and “there’s a steep learning curve,” with some of this, Geyser says. “But there’s no way of stopping it from happening. Everything is going this way.” Though EHRs might be flawed, they do represent a big leap forward in the collection of health data that could well improve the wellbeing of many seniors down the road.
From his perspective as a researcher, Walker says that he thinks the push to EHRs across the care spectrum has helped improve standards and allows clinicians to better understand how conditions tend to progress and what constitutes good care for a particular condition. It’s also helping with patient safety — continuity between care team members can be preserved and searching for a piece of information is much easier and faster when all you need to do is type a word or phrase into the computer rather than search through pages and pages of hardcopy notes.
That’s how Walker got into this field. As recently as 2011, he was pulling paper records in a community health center’s basement in Philadelphia in order to conduct an analysis of a pediatric obesity program. “I had to go into a physical record room and pull records and abstract the pieces of info I needed to do the analysis. That was extremely limiting to the kind of analysis we could do. It took a long time to pull those records. Even in 2010, it seemed like there had to be a better way.” The transformation to digital was coming, but slowly.
For many institutions, the transformation is still in process and many clinicians are still getting comfortable with whichever system they’re being told to use. One ongoing issue is that each health organization has its own system. “One of the big challenges is having EHRs talk to each other,” Geyser says. “We might use one that doesn’t talk to the hospital, which uses another that doesn’t talk to the skilled nursing facility, which doesn’t talk to the home health care agency. So, there’s all this incredible data that’s sitting in these silos that can’t communicate with one another. Once we solve that problem, it’ll be amazingly useful,” he says.
However, when a doctor or nurse becomes more focused on the documentation that’s required and loses that one-on-one communication or connection with a patient, that can be a problem. “That might change the doctor and patient or nurse-patient interaction,” he says. Because the clinician is required to write directly into the computer, there’s some trade-off there.” Medical scribes and other technologies may help fill that gap, but for the time being this is an issue in some institutions.
All that said, once that data resides in the computer, artificial intelligence tools can be applied to analyze your facility’s unique population to glean a variety of insights that could potentially result in much better care outcomes down the line. Artificial intelligence or AI is a growing field of data analysis that is changing the way many industries run, and the field of long-term care is not immune to the long arm of big data.
Walker says the term “‘artificial intelligence’ is very broad,” but usually refers to an analysis of the data to find deviations in outcome or places where data doesn’t line up with expectations. “I think one of the ways that this is being used is as a predictive model,” in that “a patient with these characteristics has a high likelihood of coming in with heart failure,” for example. This sort of tip-off of an anticipated negative consequence could help clinicians intervene earlier when the potential outcomes are more likely to be positive and less expensive to achieve.
If you’re looking to conduct a data analysis of your EHRs, Walker recommends starting out by thinking about “where are the gaps in care? What are the key questions that need to be answered for residents. In particular, is it a question of cost or quality? One of the things that the data from EHRs can help people figure out is what are some of the upstream costs of poor quality.” An analysis of where you’re spending money without adequate benefit to residents might warrant a second look or a reconfiguration of protocols or clinical approach.
But Walker cautions that even with a thorough analysis of all your records, your own data provides only a “snapshot of a piece of a large pie. Any single institution is a snapshot. You need to look at data from the primary care physicians or acute care hospital residents may also have visited to look at how they moved through these different institutions.” Claims analysis from Medicare can be a rich source of data, too, that can help researchers better understand how certain subsets of the 65-plus population tend to experience various conditions as they age. This information can help you benchmark your own facility’s performance against other similar institutions around the country in managing or treating these conditions.
Geyser says that at Inspīr, “we moved to an EHR system several years ago and it’s part of our evolution to create more sophisticated care plans. We’re also considering working with another technology provider that has a more sophisticated technology system built into the EHR that does some predictive analysis and AI in the background, so that the data that’s sitting in the system could actually detect potential changes of condition in a resident before the person has a full blown problems.”
The introduction of wearable fitness devices, such as Fitbits and other devices that track vital signs, sleep patterns, and other indicators of wellbeing, will also contribute significant amounts of data that may be useful in the future. “We’re beginning to experiment with AI in other areas as well with voice and wearable sensors in apartments,” Geyser says. These devices can integrate some of that data that’s coming from other sources and spot a problem before it happens based on subtle changes in activity levels of a resident, all without invading their privacy.
It’s early days yet and Geyser notes that “we’re just at the beginning” of exploring such technologies. But having “sensors in an apartment that can help us predict how a resident is doing or notify us if a resident is not doing well, for example, by passive monitoring of vitals, sleep, and movement patterns, can tell us a whole lot about the health state of the resident.”
Such monitoring could potentially alert clinical staff to a potential illness before the first symptoms arise, Geyser says. For example, urinary tract infections, which are common among nursing home and assisted living residents, often take a while to develop, but can cause a lot of problems once they’re full-blown. “UTIs can be really devastating, and residents may end up in the hospital with delirium, cognitive and behavioral changes, depression and anxiety, and so on. We’re working to prevent that by having wearable technology that lets us know that something’s brewing. That will prevent a lot of headaches for the family and residents, but also clinical staff and ER staff, while preventing unnecessary ER visits and hospitalizations. This is game-changing stuff,” he says, but it’s still early yet.
Walker also cautions that the old adage holds in any data analysis — “garbage in, garbage out. Medicine is as much an art as a science, and when you use AI algorithms, you have to leave room for some interpretation of what the outcome is.” It’s important to not lose sight of the human components of clinical care. Don’t lose the human amid all the data. “We can’t just trust everything to AI. You have to continue looking at what it’s telling you.”
Elaine K. Howley is a freelance journalist for various publications. An award-winning writer specializing in health, fitness, sports and history, her work has appeared in numerous print and online publications, including U.S. News, AARP.org, espnW, SWIMMER magazine and Atlas Obscura. She’s also a world-record holding marathon swimmer with a passion for animals and beer. Contact her via her website: elainekhowley.com.
Topics: Clinical , Featured Articles , Technology & IT , Wearables