AI can detect biological signs of Alzheimer’s disease years before symptoms are noted by the patient or their loved ones using only a 3-minute digital assessment delivered in a physician’s office. New research from Linus Health, an AI-driven brain health company pioneering early detection of cognitive impairment and personalized intervention, shows that this same short test can also signal cognitive impairment and flag patients who should undergo further testing with blood-based, PET, or CSF biomarkers to enable earliest disease detection when action has the greatest potential to change disease trajectory and prevent, minimize, or slow-down disability.

The findings come from two newly published peer-reviewed studies: an extensive clinical analysis of nearly 1,000 participants and an independent preclinical study conducted at Massachusetts General Hospital and the Harvard Aging Brain Study. Together, the results show that subtle patterns in behavior captured digitally and analyzed by AI can reveal early disease processes linked to amyloid and tau deposition in the brain, long before traditional tools detect change.

“This is the first demonstration of a true behavioral surrogate biomarker for Alzheimer’s disease,” said Linus Health CEO and Co-Founder David Bates, PhD. “AI is detecting the earliest disruptions in brain function, likely several years before clinical symptoms. These groundbreaking studies confirm that our AI-driven assessments offer reliable preclinical signals of emerging pathology that can help clinicians act at a time when action matters most.” 

A First-Ever Demonstration of AI ‘Multiplexing’ in Brain Health
The first study, Concurrent detection of cognitive impairment and amyloid positivity with a multimodal machine learning-enabled digital cognitive assessment, published in Alzheimer's Research & Therapy, demonstrates that Linus Health’s Digital Clock and Recall (DCR) assessment can generate multiple clinically meaningful predictions from the same 3-minute task. This represents the first demonstration of what researchers call AI “multiplexing” in brain health: the ability to use a single stream of behavioural data to power separate AI models that each deliver a distinct clinical insight.

In this study, multimodal sensor data, including drawing behaviour, moment-to-moment pauses, speech characteristics, and signals from a digital stylus, were analysed to estimate cognitive impairment and the likelihood of amyloid positivity on a PET scan. Across nearly 1,000 participants, the DCR showed strong accuracy for both outcomes, performing on par with or better than traditional tests and complementing Alzheimer’s blood tests to increase their predictive performance.

A brief DCR assessment can therefore help physicians determine which patients may benefit from an Alzheimer’s blood test, a specialist referral, or closer monitoring, all based on a single, brief assessment performed in an ambulatory setting. 

Independent Study Shows AI Detecting Alzheimer’s Biology in Asymptomatic Adults
A second study, Higher Amyloid and Tau Burden Is Associated With Faster Decline on a Digital Cognitive Test, published in Annals of Clinical and Translational Neurology, showed that Linus Health’s Digital Clock Drawing Test (DCTclock) predicted baseline levels and longitudinal changes associated with amyloid and tau in older adults who had no symptoms of impairment. In a longitudinal study of more than 200 cognitively normal older adults conducted by Massachusetts General Hospital and the Harvard Ageing Brain Study team, those with higher levels of Alzheimer’s-related amyloid or tau on PET brain scans showed faster decline on the DCTclock—especially on measures linked to information processing speed and executive function. 

These results reinforce a key scientific finding: AI-based analysis of process metrics of everyday behavior can capture early disruptions in brain function that can identify Alzheimer’s pathology. The study also showed that these early signals are found long before traditional cognitive tests detect change.

The two studies offer converging evidence that easy-to-use digital tools can reveal early changes in brain function at scale, using assessments that fit naturally into primary care workflows. This comes at a time when specialist shortages, long wait times, and rising demand for cognitive evaluation make early access increasingly difficult.

“Patients need answers long before the 9- to 18-month wait many families face today,” said Linus Health Chief Operating Officer John Showalter, MD, MSIS. “AI-based tools can help primary care teams surface the right concerns earlier, direct patients to the right care sooner, and support timely decision-making for families.”

These findings are only the beginning. The groundbreaking research validates the direction Linus Health is pursuing, with one assessment producing multiple insights, expanded risk modeling that incorporates symptoms and biomarkers, and more sensitive detection of early cognitive signals. 

“We are moving toward a future where AI helps clinicians see what has always been hidden, far earlier than previous tools ever allowed,” Bates said. “This is the foundation for proactive brain health where we can identify issues early, support individuals earlier, and help clinicians guide the path forward when it has the greatest potential to improve outcomes.”