Use of artificial intelligence (AI) has been steadily increasing at nursing homes, as the technology begins to deliver on its promises to improve efficiencies in a wide variety of areas, from clinical outcomes to back-office work.
And while ChatGPT is the attention-grabber these days, natural language processing is but one of the ways AI is benefitting the SNF world.
From clinical decision support, to improving staff engagement and the resident experience, to assessing risk for falls and pressure ulcers, to having robots serve food and clean, AI-powered tools are making their way into nearly every facet of nursing home operations. And while the technology is still developing — and the risks and limits of AI are real — the tech is helping drive improvements in care while addressing bottom lines and staffing challenges.
Lisa Chubb, chief clinical officer at Brickyard Health, said technological advances and AI tools will be driving some of the changes that her organization has incorporated as a result of the pandemic, with a big goal being to free up nurses to spend more time with residents.
“We’ve got to provide cost-effective quality of care. So always having those predictive tools and analytics, so that you can be very quick to change the clinical decision making and affect your outcomes, is very impactful right now for Brickyard,” Chubb said.
In recent years, Brickyard, like many other nursing home providers, has been impacted by regulatory and compliance changes. These have increased the administrative burden on nurses, but AI tools for electronic health management and patient monitoring have helped Chubb’s organization tremendously in this regard as well.
“If you build your electronic health record around the regulatory requirements and all of the data that you need to capture to actually show and capture the burden of care in the most efficient way possible, that’s how you can be successful,” Chubb said. “And really, that’s what it boils down to, and it also allows the nurse to be closer to the bedside and less in front of the device.”
AI to battle common conditions in SNFs
AI’s assistance with clinical decision-making has meant improvements in the accuracy and power of diagnosing and treating disease, in turn leading to better clinical outcomes and more time spent with patients, according to Majd Alwan, a digital health and aging services technology expert who currently serves as ThriveWell Tech’s chief strategy and growth officer.
Pressure ulcers, wound management, hypertension, congestive heart failure (CHF) and diabetes are some of the more common conditions in nursing homes, and AI tools are being used to manage these diseases and more, Alwan said.
AI will also enable remote models of care – an area that is currently getting attention for allowing staff flexibility.
“Another area of clinical decision support systems will be telehealth and biometric remote patient monitoring … that warn clinicians when biometrics are out of line,” said Alwan. “[AI] clinical decision support systems are geared for handling multiple chronic conditions and comorbidities.”
Alwan told Skilled Nursing News that he also sees AI as addressing certain racial health disparities seen in treatment and diagnosis of common health conditions found in nursing homes.
Predicting and diagnosing pressure ulcers – a condition often missed on darker skins due to being difficult to visually detect in early stages of development – is one such example, said Alwan, who has observed the wide successes of AI at Asbury Communities, a leading not-for-profit system of continuing care retirement communities (CCRCs), of which ThriveWell is a subsidiary.
“All the nurse needs to do now is connect a camera to a computer and take a picture of the patient’s back, and it will identify exactly where the pressure points are,” said Alwan of the AI tools used to detect pressure ulcers. “You are preventing [a pressure ulcer] from developing into stage one, and maybe even stage two pressure ulcer which would be much, much harder and costlier to treat.”
Such AI-enabled tools are also being used to manage CHF patients.
For CHF patients, a simple weight and blood pressure check can now alert the clinical staff to modifications in crucial medications. Lasix removes excess fluid and treats CHF as well as kidney disorders and liver disease, but its use requires tedious monitoring of the patients. With AI’s guidance, this process is less laborious – and even more accurate.
“If the system sees a sudden change in [the patient’s] weight, the system may ask them if they’re taking their Lasix, and if they’re not right, it may give them an education about the importance of medication adherence and taking their Lasix,” Alwan said. “If they complain, it may ask them why they’re not taking their medication, and if they say, ‘The frequency of nightly bathroom visits is disrupting my sleep,’ then the clinician may come back and give them recommendations to change the schedule of taking their Lasix.”
AI’s ability to enable better ways for disease detection, diagnosis, treatment and management, including for medication, will go a long way towards improving not only quality of care at SNFs but will reduce costs too, believes Alwan.
How ChatGPT is working
The applications of AI extend beyond clinical uses within Asbury Communities, Alwan said. AI tools can support business decision making as well.
“You could be looking at occupancy data, you could be looking at market data, readmissions, referrals, dependent forms, even payer data, claims data, and if you have the right data, you could be making the right inferences and insights,” Alwan said.
Hospitality services within nursing homes are using AI tools too.
“It’s not just software and robotic process automation, but physical robots that we’re starting to see in dining, bussing, delivery, disinfection and cleaning, as well as lawn mowing,” Alwan said.
But the most prevalent application of AI is in the area of natural language processing.
“Voice control devices are being used as intuitive interfaces for a lot of software packages, and that’s more resident facing … it’s making residents’ lives easier,” Alwan said.
And in this regard too, the much talked about ChatGPT is great for answering questions and crunching information and providing clinical support, Alwan said.
“[ChatGPT] is not a replacement for clinicians’ expert opinion. But again, it can allow staff to perform and sort of have knowledge at the top of their license, and it allows them to potentially detect and find issues and elevate those issues to a supervising clinician, like a physician and medical director,” said Alwan. A popular category of clinical questions is regarding rare medical conditions, for example.
That said, ChatGPT touches only the surface of what AI can do in the SNF and health care space, Alwan said.
“ChatGPT is only one instance of a deep learning AI that has now come into the public domain,” Alwan said.
Moreover, AI is being used for medication management as well. This endeavor can be a difficult task considering nursing home patients are prescribed so many medications at a time. Side effects from interaction of different pills can be difficult to track, but here too AI is seeing widespread use.
“If that new prescription contraindicates another medication that the patient or resident is already on, [an AI tool] will flag it. It will identify risk factors and prevent medication errors and adverse drug reactions,” Alwan explained.
However, communication remains one of AI’s most effective uses today in the senior living, according to some experts.
For Peter Kress, CIO at Acts Retirement-Life Communities and Director of Information Technology at Willow Valley Communities, the best use of AI in senior living will come from communication and personalizing language to improve customer experience and engagement.
“At Acts, we’re trying to reinvent what wellbeing means for our residents, and our residents who’ve been healthy and independent,” said Kress. “Is someone more fit? Are they sleeping better? Are they enjoying the food more? Are they having a better social engagement with the people around them? Are they living longer and better? Are they living happier? Are they living smarter? And, frankly, we see AI as an essential amplifier of the strategies and programs that we put in place to help our residents thrive,” said Kress.
AI use for staffing solutions
Staffing challenges facing the industry have led to AI innovations to tackle workflows and screen, recruit and retain workers.
AI staffing solutions include those pertaining to pulse surveys and other ways to engage employees better. The vast majority of new hires who enter senior living leave in the first 90 days, which makes it imperative to find ways to prolong their careers at a facility, Alwan said.
“We firmly believe that AI has a great potential in engaging candidates, understanding their issues, whether it’s related to the nature of the work, whether it’s related to supervision, training and so on,” he said.
AI applications that allow for flexibility in scheduling and the ability to self-schedule, along with pay differentials based on difficulty of shifts, promise to alleviate staffing issues.
“Some of these [AI tools] learn the patterns and the times that are preferred by employees … some of the applications are going to be extremely, extremely relevant and important given the significant shortage of caregivers that we see today,” Alwan said.
Aside from schedule workarounds, the applications ease and enhance a range of HR tasks such as candidate screening based on automatic scanning of submitted resumes for keywords of the job description and the skills and experience listed by the applicant on the resume. This AI tool short lists and flags potential candidates for the HR department.
And while AI can’t increase the number of candidates, it can nab them before they decide to enter other sectors by allowing HR to act in a speedy fashion.
There are applications that utilize chatbots to automatically schedule an interview with a selected candidate by having access to their calendar and contact information – with the appearance that it’s coming from the HR department.
“And that not only engages the candidates, but shortens the time for the first interview in the hiring cycle, in turn increasing the chances of getting candidates hired before they accept another competing job offer from another industry,” Alwan explained.
Limits of AI
Despite all the efficiencies and assistance AI tools can help with, Alwan cautions that error can seep into applications because of the data used to build these applications.
“One of the biggest problems that AI may face is the fidelity of the data,” Alwan said. “So there could be missing data, there could be erroneous data, there could be discrepancy in the data of the resident from one system to another for a multitude of reasons.”
And so, for the “predictive modeling” to be error free and useful for making the right decisions whether it’s for clinical decision support or business decision making, the data has to be accurate.
“If the data is corrupt, as the saying goes, ‘It’s garbage in, garbage out.’ So that is the biggest risk, in my opinion,” Alwan said.