Background: Scalable early detection for cognitive impairment diseases (CID) is urgently needed. Current tools are polarised: cognitive tests lack sensitivity and objectivity, while gold-standard biomarkers (plasma p-tau217, amyloid-PET) are inaccessible for population screening. A rapid, objective, biologically-informed point-of-care assessment is required.

Methods: In 180 individuals (60 cognitively normal, 60 mild cognitive impairment [MCI], 60 Alzheimer's disease [AD]), we assessed a 1-minute digital finger movement (FM) task via high-precision magnetic sensor, extracting 248 kinematic features. These were integrated with multimodal data (plasma Aβ42/40, p-tau181, p-tau217, NfL, GFAP; neuropsychology; amyloid-PET). LASSO regression identified key features.

Results: Bimanual alternating FM features (e.g., IL_dist1, IL_dist3) correlated moderately with AD pathologies (p-tau217, amyloid-PET) and strongly with cognition. This signature accurately identified AD (AUC 0.886 [95% CI 0.818–0.954]), MCI (0.816 [0.741–0.890]), and any cognitive impairment (0.738 [0.635–0.842]) with 8333% sensitivity and 86.67% specificity. The test requires no laboratory, specialist, or blood, providing real-time results.

Conclusion: This 1-minute digital FM assessment is a novel, biologically-validated behavioural biomarker for AD, bridging molecular pathology and clinical symptoms with an accessible, objective tool. It offers a scalable, pathologically-grounded screening solution with immediate potential to transform early detection in primary care and democratise access to early intervention worldwide.