Primary care is often the first point of contact for patients with cognitive complaints, making initial cognitive screening an essential step to avoid delays in diagnosing Alzheimer's disease (AD) at an early stage. We developed MemScreen, a self-administered smartphone application that assesses overall cognition and verbal memory, and evaluated its ability to detect mild cognitive impairment (MCI) in both general and clinical populations.

We conducted two validation cohort studies: (1) UK-based Whitehall II cohort study (13^th wave, 2018-2022) involving a general population (MCI defined by poor performance on a global cognitive score), and (2) five French memory clinics involving patients without dementia (amnestic MCI defined by the Free and Cued Selective Reminding Test). MemScreen, MMSE, and TMT-A effectiveness was assessed using Area Under the Curve (AUC) values from unadjusted and adjusted logistic regression models.

In Whitehall II (n=2118, mean age 75.9 years, 23.9% women, 14.5% MCI), median MemScreen completion time was 4 minutes 18 seconds. MemScreen had the highest AUC (0.87; 95% CI: 0.82-0.89) for distinguishing MCI, outperforming MMSE (AUC=0.79; 0.76-0.82; p=0.018) and TMT-A (AUC=0.77; 0.74-0.80; p=0.023). MemScreen sensitivity and specificity were 78.6% and 78.7%, respectively. In memory clinics (n=303, mean age 70.5 years, 53% women, 46.9% amnestic MCI), median completion time was 5 minutes 17 seconds. MemScreen showed superior performance (AUC=0.87; 0.83-0.91) compared to MMSE (AUC=0.72; 0.67-0.78; p<0.001) and TMT-A (AUC=0.63; 0.56-0.69; p<0.001), with 93.0% sensitivity and 54.0% specificity for amnestic MCI.

MemScreen outperformed traditional tests in identifying MCI in both general and clinical populations. Its self-administration and short completion time suggest potential as an effective screening tool to optimize memory clinic referrals for AD diagnosis and treatment.