A new artificial intelligence (AI) model can predict a future onset of Alzheimer’s disease through analyzing how people describe a picture of a cookie theft.

The study, conducted by CES® veteran and Consumer Technology Association (CTA)® member IBM in partnership with Pfizer, finds that the model can predict with 71% accuracy that someone will develop Alzheimer’s disease, even seven years before a clinical diagnosis is made.

Catching the Problem Before It’s a Problem

Early intervention strategies can decrease the risk, delay the onset or slow the progression of Alzheimer’s disease. According to the study, because cognitive decline associated with Alzheimer’s disease often manifests in agraphia and other language comprehension problems, linguistic abilities can — to an extent — be utilized as a prognostic biomarker of preclinical Alzheimer’s.

The cookie-theft picture-description task used in the study is often used to evaluate cognitive impairment for a variety of neurological disorders.

IBM and Pfizer’s work differs from studies before it largely because data was collected from subjects who showed no signs of cognitive impairment. The team also evaluated subjects in the general population instead of those in high-risk groups.

AI Dissects the Descriptions

The artificial intelligence (AI) model was able to pull data from digital transcriptions of the description task and detect linguistic features sometimes related to early signs of cognitive decline.

The model took into consideration more than 87 variables, including punctuation, verbosity, repetition and misspellings. Paired with data on age, gender, education, attention and other factors, the AI model could analyze and find patterns to predict the future clinical diagnosis of Alzheimer’s.

Why It Matters

“Ultimately, we hope this research can lead to the development of a simple, straightforward and accessible metric to help clinicians assess the risk of Alzheimer’s disease in an individual, leading to earlier intervention,” said Guillermo Cecchi, IBM’s principal researcher for computational psychiatry and neuroimaging.

The team also hopes to leverage the AI model technology to better understand diseases such as schizophrenia and Parkinson’s disease.