The AI-Powered Test That Could Revolutionize Dementia Detection

"The analogy I like to use is that of a blood pressure test," explained Dr. Thomas Sawyer, chief operating officer at Cognetivity, a healthcare startup building an AI-powered test to detect the early signs of dementia.

By Anna Codrea-Rado, contributor

“The analogy I like to use is that of a blood pressure test,” explained Dr. Thomas Sawyer, chief operating officer at Cognetivity, a healthcare startup building an AI-powered test to detect the early signs of dementia. “It’s a general look into your cognitive health.”

The company is currently seeking FDA approval for its Integrated Cognitive Assessment (ICA) to be used as an aid in detecting dementia 10-15 years before visible signs would commonly be recognized.

The ICA, which can be administered on a tablet device, works similarly to a CAPTHCA test (those tests you might encounter when setting up a new email account or purchasing tickets that quiz you on whether you are a human or a robot), except the images move in rapid succession. Patients taking the ICA are asked to indicate as quickly as possible if they see a specific type of photo. For example, patients might be asked to watch for pictures of animals in a selection of nature shots. Behind the scenes, AI algorithms are clustering the test performance to give a measure of the patient’s cognitive health.

“A blood pressure test is a reliable, cheap indicator of a cardiovascular issue,” said Dr. Saywer. “The same goes for the ICA. If you go to your doctor for a regular check-up, you can do this test and report into the doctor’s healthcare system.”

Dementia Diagnosis 2.0

Catching dementia early can have life-changing results, but existing diagnostic tools are lacking. The degenerative disease currently affects 50 million people worldwide according to the World Health Organization, and will affect 1 in 3 children born today. According to the Alzheimer’s Association, the cost of treatment is expected to reach $1.1 trillion dollars by 2050.

While there’s no cure for the disease, early diagnosis is vital, as it allows a patient to better prepare, take symptom-lessening medication, and delay entering residential care. The Alzheimer’s Association estimates that early and accurate diagnosis could save up to $7.9 trillion dollars in medical and care costs.

Getting a diagnosis early enough, however, is complex. There’s no one test to take; rather, patients need to undergo a battery of tests, including a full medical history, scans, and memory tests.

These tests have serious limitations. A recent study published by the American Academy of Neurology found that these tests can be inaccurate. In the study of over 800 Americans, researchers found that more than one-third of the patients were misclassified by the tests.

“Dementia can be difficult to accurately detect, particularly in a primary care setting,” noted study lead author and dementia researcher Janice Ranson. “Our results suggest that some of the misclassification is due to test biases, such as a patient’s age, ethnicity, or education level.”

Ranson’s research points to a need for data science to improve the shortcomings of these existing early assessments—a gap the AI-driven ICA is trying to fill. “We desperately need more accurate and less biased ways of detecting dementia swiftly in the clinic,” Ranson said.

Big Brain Data

Cognetivity’s test aims to overcome the limitations of other screening assessments through its new diagnostic technology and the cluster analysis underpinning it. Existing tests, such as the Mini-Mental State Examination (MMSE), sometimes under-diagnose highly-educated patients. But because it’s a rapid visual categorization test, a patient’s education level has no bearing on the ICA’s results, and a patient can’t learn how to take it.

Where patients’ demographic data, such as age and gender, could influence their likelihood of cognitive impairment the AI maps this accordingly. The ICA uses deep learning to determine patients’ results in comparison to learned clusters. The AI will be able to tell a healthcare provider what a patient’s cognitive profile should look like based on these predictors; the more the test is used, the more sensitive it becomes.

The data Cognetivity is able to collect has additional benefits to researchers of other screenings. “Once we’ve looked into some detail of large population data, we believe there will become relationships between the speed and accuracy of different image compositions, which may be characteristics of different diseases,” Dr. Sawyer said.

Sawyer sees a future in which tests like the ICA will not only be able to detect if someone is at risk of dementia, but pinpoint what type of cognitive impairment he or she is presenting. For now, though, the company has its sights set on its simple test becoming a regular feature in the doctor’s office—something as routine as a blood pressure test.