AI-Based Cognitive Assessments Help Diagnose High-IQ Patient with Alzheimer Disease

An 82-year-old woman presented to Cleveland Clinic from out of state with short-term memory impairment, depression and anxiety. Despite undergoing two recent cognitive tests that resulted in normal findings, the woman and her husband returned for a third test. When Cleveland Clinic geriatrician Ardeshir Hashmi, MD, met with the patient and her husband, they explained that cognitive impairment was indeed affecting the woman, despite normal test results.

High IQs can result in higher test scores

During her career, the woman worked as a human resources executive at a Fortune 500 company.

“Patients with high IQs, as in this case, are often able to rely on their high cognitive reserve during the standard, traditional, paper-and-pen cognitive test,” Dr. Hashmi, who is Endowed Chair for Geriatric Innovation and directs the Center for Geriatric Medicine, explains. “Essentially, she was outsmarting the standard test and her memory impairment deficits weren’t showing up.”

While existing tests like the Montreal Cognitive Assessment (MoCA) and Mini Mental State Exam (MMSE) are the current pen-and-paper based standards for screening, challenges like the “ceiling effect” and inter-tester variability accompany these tools.

Machine learning algorithm gives more accurate results

Instead of utilizing the standard testing for a third time, Dr. Hashmi directed the 82-year-old to use an AI-based cognitive assessment on a self-administered computerized platform. This machine learning algorithm consists of a complex memory test that can be completed in about ten minutes.

“The AI-based cognitive assessment showed significant memory deficits and indicated that Alzheimer’s disease is, indeed, happening here. It was just never picked up by the standard test,” Dr. Hashmi says. “Because we now know this information, we can prescribe medications that will stabilize her cognitive abilities and help prevent future decline.”

Dr. Hashmi plans to have the patient repeat the machine learning test in three months to ensure the medications are stabilizing her cognitive abilities. “If she scores the same, that’s a success for us all.”

Multidisciplinary care, dynamic team

Dr. Hashmi credits the multidisciplinary team, especially the behavioral and cognitive therapists, at Cleveland Clinic for their collaborative teamwork.

“We work with a very dynamic team. Care is tailored for each patient,” Dr. Hashmi says. He mentions that visits within the Successful Aging Program at the Center for Geriatric Medicine, unlike many appointments with a primary care physician, allow the luxury of time. “An entire team of individuals with different areas of expertise and knowledge spend an hour talking with a patient and his or her family.”

The 82-year-old plans to continue cognitive therapy at her local clinic in her home state.

“Medication is responsible for about two percent of retaining cognitive abilities,” Dr. Hashmi says. “Ninety-eight percent of maintaining that capacity comes from brain exercises, word games, puzzles, online exercises and music.”

Researching automated screenings is not a new endeavor for Dr. Hashmi. In 2019, he conducted a study with colleagues from the University of Massachusetts to test a machine learning algorithm that analyzed cognitive abilities. The findings were reported in the Journal of Geriatric Psychiatry and Neurology.

Earlier detection means more options

Recently, the National Institute on Aging awarded Cleveland Clinic a grant that will help researchers continue developing and testing automated screening tools.

Dr. Hashmi hopes these tools will help identify memory changes sooner and more accurately than the traditional paper-and-pen testing methods.

“We meet these people way too late in the game,” Dr. Hashmi says, noting that memory impairment can go unnoticed over a span of 20 years. “By the time we’re meeting with patients, options are limited. Medications can help stabilize further decline, but unfortunately can’t reverse what’s already been lost. If we can help identify impairment earlier and start medications sooner, we can slow the disease. That buys patients time with their families and allows them to set up necessary resources for the future.”