Features

How AI, Digital Screening Tools Can Help Flag Early Cognitive Decline

A new pilot program is leveraging AI-based digital screening tools to drive early detection of cognitive impairment in the primary care setting.

Source: Getty Images

- Alzheimer's disease and Alzheimer's disease-related dementias (AD/ADRD) present a challenge for clinicians as the disease burden and toll on individuals, caregivers, and health systems rise.

Currently, the National Institute of Neurological Disorders and Stroke (NINDS) at the National Institutes of Health (NIH) estimates that dementia affects over 6 million people in the US and 55 million worldwide. There are currently no known treatments to prevent or stop the progression of the disease.

Early diagnosis of Alzheimer's and other dementias remains at the forefront of efforts to minimize the impact of these neurodegenerative diseases. But challenges such as increased life expectancy and the risks of aging, along with complexities in diagnosis and treatment resulting from mixed brain pathologies, make early detection difficult.

Tools to address these challenges, such as blood diagnostics, predictive analytics models, and frameworks to differentiate Alzheimer’s from other dementias, have been developed, and some are used in the clinical setting. But these cannot address some of the biggest hurdles to early detection: care access for patients and training for primary care physicians.

A new pilot program at Indiana University School of Medicine and Indiana University Health, in collaboration with the Davos Alzheimer’s Collaborative (DAC), seeks to tackle these issues using digital screening tools.

Jared Brosch, MD, a neurologist at IU Health and assistant professor of clinical neurology at IU School of Medicine, and Phyllis Ferrell, global head of external engagement for Alzheimer’s disease at Eli Lilly & Company and director of the DAC Healthcare System Preparedness initiative, recently spoke with HealthITAnalytics about the pilot as a model for shifting cognitive care from reactive to proactive.

THE IMPORTANCE OF EARLY COGNITIVE IMPAIRMENT DETECTION

Enabling early detection, generating actionable insights for interventions, and identifying patients that might benefit from emerging therapies are key aspects of improving ADRD outcomes. But these rely on widespread screening and an accurate understanding of what a screening should look for: signs of mild cognitive impairment that appear long before the disease.

“Dementia is a very late stage [of disease] that often begins with mild cognitive impairment,” Ferrell explained. “Typically, dementia means that you've got a significant functional impairment. Often with [ADRD], you get this mental model of someone who's very, very late in their disease.”

Dementia is the latent symptomatology present at this disease stage, so the pilot project aims to detect cognitive impairment early to help delay the onset of late-stage disease. However, she noted that the current clinical approach means that most people are identified once they reach the later stages of ADRD.

“And yet, we know that cognitive impairment and some of the symptoms can be caught a lot earlier,” Ferrell stated. “We know now that the pathology of Alzheimer's disease actually starts in the brain 10 to 20 years before symptoms even start. But we used to not be able to see that because you could only see it in a living brain, and there weren't a lot of people that were alive that volunteered for brain biopsies. So, you'd have to actually get a brain biopsy or do an autopsy.”

Now, innovations in medical imaging, blood testing, and other biomarker-based diagnostics allow clinicians to observe the pathology in the brain that indicates the presence of disease years before symptoms arise.

Despite these advances, however, many challenges still prevent the early detection of cognitive impairment in the clinical setting, Brosch pointed out. He explained that cognitive function and impairment are on a continuum.

However, fear of a potential ADRD diagnosis and resistance to screening are major roadblocks clinicians face when promoting screening and early detection.

“People don't necessarily want to have things detected early,” Brosch explained. “They're scared of [the results]. They're like, ‘Well, is this just a normal aging process?’ or ‘Maybe I shouldn't be worried about this,” or ‘I definitely don't want to get what mom or dad had, so I'm just not going to say anything about it.’”

These fears can create a barrier that prevents patients from going to their doctor with complaints they may have related to their cognitive function, which can lead to further cognitive impairment and ADRD only being detected once it’s too late to address effectively.

Another challenge lies within the healthcare system itself, and how it is set up, Brosch stated.

“[Clinicians] are under a lot of time pressures to see people, to address a specific problem, and then to move on and see a lot more people because there's not enough providers, and there's a lot of people who need help,” he said, further explaining that current efforts to combat this issue aren’t sufficient.

“Medicare is trying to help through what's called the ‘Annual Wellness Visit’ for its Medicare recipients. So, we're talking about people who are over the age of 65, and this doesn't necessarily help the people who are developing early onset dementias,” he added.

Ferrell echoed this, noting that the healthcare system has not evolved at the same pace as ADRD advancements.

“The system we have today is designed for the treatments that we've had for the last 30 years, and 30 years ago, the only thing we could do was just wait and see to see if something was really Alzheimer's disease,” she explained. “I think the challenge we have now is we've got [newer] innovations that allow us to diagnose more accurately because we can now see the disease in a living brain.”

There are also some treatments and lifestyle modifications that can have a significant impact on patient outcomes and overall brain health if started sooner rather than later.

“There are things we can do for brain health today. So, this whole concept of ‘I don't want to talk to anybody about [my concerns] because there's nothing I can do’ isn't the case anymore, but the system is still designed for that,” Ferrell said.

One way to promote a more proactive approach to these issues is to make screening effective and accessible in the primary care setting.

LEVERAGING DIGITAL SCREENING TOOLS IN PRIMARY CARE

Through the pilot project, IU and DAC aim to bolster early detection of cognitive impairment in primary care through clinician education and digital screening tools.

Brosch indicated that showing primary care physicians why early detection is critical, what tools they can use to facilitate it, and which treatments and lifestyle modifications may help patients with cognitive decline is a major component of the pilot. From there, clinicians can utilize digital screening to assess patients during routine care.

“What we've done is we've digitized and allowed the primary care doctors to use a tablet that they hand to their patients when they come into the office, and they say, ‘Hey, we're doing this quick cognitive screening test. It'll only take a few minutes. Just follow the instructions on this tablet,’” Brosch explained.

The test itself, the Linus Health Core Cognitive Evaluation, is comprised of a lifestyle-based questionnaire and a digital cognitive assessment, which asks patients to remember three unrelated words, complete the clock drawing test, and repeat the three words. The tool then uses artificial intelligence (AI) to detect subtle signs of cognitive impairment and generates a score.

“[Using that score,] they get a stoplight that's green, yellow, or red. And the doctor can take a quick look at it and decide, ‘what do I do with this?’” Brosch stated. “'This person's a green; they're good to go. I'll talk to them about their lifestyle,’ or ‘It's yellow; maybe I should refer them to someone to do a little bit of a deeper dive to figure out if there really is an issue here,’ or if it's red, ‘Maybe I should refer them directly to one of our experts for additional testing, diagnosis, and treatment.’”

He explained that the digital tool helps analyze more nuances that may correlate to cognitive decline that the traditional pen and paper approach can’t show, such as how long it took the patient to draw the clock, the shape of their drawn clock compared to a circle, how well their numbers were placed, and what do their hands looked like while completing the test.

According to Ferrell, this approach not only allows cognitive impairment to potentially be detected earlier but could also help alleviate resource challenges. As the population ages and lives longer, the demand for cognitive care grows, but the number of neurologists and other specialists may not grow similarly.

“We have a rapidly aging population, so we need to do things differently as we have more people turning 65,” Ferrell said. “You can't keep pushing those people into our most expensive resources.”

Brosch currently has a long wait list of patients needing to be seen, like many other specialists in the field.

“So, the beauty of these digital cognitive assessments is they allow for what we call task shifting, which means you can take a task out of a more expensive resource and put it into a less expensive resource,” Ferrell explained. “And that doesn't mean that those folks won't eventually get to [Brosch]. It just means that the ones that really need to get there will get there, and we can reassure those that are just fine and probably don't need to go into those further and more expensive assessments.”

This task shifting is a key component for improving patient flow, but there is more to it than simply giving clinicians new tools. Infrastructure changes are also needed, Brosch stated.

“We are a larger system, and that has allowed us to do a few things from an infrastructure standpoint to make this successful,” he said. “And one of those things is we have a nursing role that is completely new. We call this nurse the Brain Health Navigator.”

“If [a patient is] a yellow or a red on their stoplight, they are, automatically in some cases, or subjectively, based on that doctor's preference, sent to this nurse who's trained to do some more in-depth neuropsychological tests. So again, we're task shifting, and we're having this person do a deeper dive that doesn't involve having to wait to go see a neurologist, a geriatrician, or a neuropsychologist,” Brosch continued.

This helps expedite the process and funnel patients so that they can be seen as quickly as possible in the most appropriate clinical setting.

However, there are various barriers to leveraging these cognitive assessments, whether they are digital or not, Ferrell noted.

One that applies specifically in the context of the pilot is that screening patients with these new tools isn’t what many clinicians are used to doing. Additionally, patients often don’t want to discuss cognitive decline or why those conversations are important. This is where the education component of the project comes in.

But other, less obvious factors also play a role.

“There's barriers around billing codes and things that should be really easy to fix, but they're not,” Ferrell explained. “[These issues are] about how these tests get reimbursed and how to make sure that we don't spend more of a doctor’s time than we need to make sure that these screenings get done.”

The pilot project at IU is helping to generate data that can be presented to the US Department of Health & Human Services (HHS) and the Centers for Medicare & Medicaid Services (CMS) to help develop actionable solutions.

But some of the other barriers are more health system- or clinician-specific. To combat these, Ferrell and Brosch described how the pilot’s approach can be adapted.

ADAPTING IMPLEMENTATION TO SUPPORT DIVERSE CLINICAL WORKFLOWS

The project uses an agile implementation framework, an approach that focuses on being flexible and continually adapting to new information and inputs from those participating in the pilot. By leveraging an agile implementation approach, those running the pilot can gain real-time insights and ensure the project evolves over time.

“As a pilot project, each clinic using agile implementation is also customizing their own workflow,” Brosch explained. “So we’ve found, and we believe, that one workflow doesn't fit all, and we want our physicians who are in charge of their clinics to really have a say in how [the tool] gets implemented.”

By allowing clinicians to decide whether the tool gets used at the beginning or end of a patient visit or whether the test is given by a medical assistant or the clinician themselves, the agile implementation approach has helped bolster clinician acceptance and buy-in without disrupting clinical workflows, he indicated.

“One thing we always say is when you've seen one health system, you've seen one health system,” Ferrell stated, noting that IU is implementing the pilot within six of its clinical settings, and each has found a slightly different workflow most effective.

“I think that's super important for us as we think about solutions,” she continued. “How do you give people the parts of the solution that are mandatory, that have to be fulfilled? And then where do you allow for local adaptation so that different systems, different settings can customize what works for them?”

Thus, program leaders are not the only ones responsible for adapting as the pilot progresses. Rather, the clinicians at each implementation site have had a role in shaping the program since the pilot’s launch. Instead of handing clinicians a pre-determined implementation protocol, care teams were asked to help co-design the project.

“[Agile implementation of the pilot] is a two-way street,” Brosch said. “It's not just the physicians needing to adapt. It's the physicians who are providing feedback and allowing for modification of the software, and we've found that Linus has been able to improve the reliability of their software through this real-life implementation.”

Using the software, clinicians are gaining some unexpected insights into their patients' needs and wants. Often, clinicians are surprised when patients they’ve known for some time score in the red, which Ferrell stated speaks to the sensitivity of standardized digital cognitive assessments when combined with clinicians’ observations about their patients.

“We've got a couple of sites that are in very affluent areas, one of which where we only see people age 65 and up,” Brosch explained. “We also have some that are in very low-income, low-education areas, and it's interesting to see that if you get a yellow on the stoplight in a low-education, low-income area, it may be because that person had poor hearing, or they didn't understand the word, or they didn't understand the instructions.”

The physician can then weigh in on that result based on their knowledge of the patient, which can aid clinical decision-making about how to proceed with different patients, he added.

Another finding that the project has highlighted is that some patients are not as resistant to conversations and screenings around cognitive impairment as clinicians expected them to be, which Ferrell pointed out may indicate that patients value knowing information about their brain health if they feel there is something they can do to address it.

As the pilot continues, Brosch and Ferrell hope that these insights, alongside feedback from clinicians, can make digital tools faster and more efficient and help significantly tackle the burden of AD/ADRD.

“We're really excited to be a part of this project because if you look at the research community when it comes to Alzheimer's and dementia, it's all about detecting early, treating early, and those are really, really the absolute keys in all of this,” Brosch said. “If we can help develop tools, and even if a pilot is a failure, which I don't think it will be, but even if it is, we are learning something about early detection and how to modify our healthcare system to be even better.”