Introduction Early detection and prognostic prediction of Alzheimer’s disease (AD) is essential for timely intervention and improved patient outcomes. However, current diagnostic methods, including cerebrospinal fluid (CSF) analysis and neuroimaging techniques, are often invasive, costly, and unsuitable for early screenings. Non-invasive neural recordings like electro- or magnetoencephalography (EEG or MEG) provide a non-invasive alternative, yet they often struggle to identify cortical alterations associated with AD at preclinical stages. To address these limitations, we propose a novel approach based on digital twin models that extract personalized digital biomarkers reflecting individual neurodegeneration levels from non-invasive neural recordings.
Methods We developed the DADD (Digital Alzheimer’s Disease Diagnosis) digital twin model to derive digital biomarkers from non-invasive neural recordings [1, 2]. DADD reconstructs personalized levels of neurodegeneration from individual neural recordings via biophysical modeling of AD-related functional alterations. DADD parameters reconstruct patient-specific levels of synaptic degeneration, network-level brain disconnection and neuroplastic rewiring. DADD biomarkers were used to predict evolution of cognitive decline in 459 participants with prodromal AD, also testing their cross-center and cross-modal generalizability on additional two cohorts totaling 46 HC and 76 MCI participants undergoing different types of neural recordings.
Results
The DADD model significantly outperformed standard EEG analysis in predicting follow-up results in concurrent machine-learning classifications (AUC=0.71 vs AUC=0.62, p<0.00001). Combining DADD digital biomarkers with CSF biomarkers significantly increased the prediction of future conversions (AUC=0.81 vs AUC=0.74, p=0.009). A Cox Proportional Hazard model found that digital biomarkers had the highest predictive power for future conversions in SCD participants (HR=1.94, p=0.003), also outperforming CSF biomarkers (HR=1.40, p=0.013). In the cross-center classification, digital biomarkers obtained consistent cross-center classification (77–78% accuracy), while standard biomarkers performed poorly in the generalization attempt (56–65%) [3].
Discussion These findings suggest that DADD might be a powerful tool for early AD diagnosis and prognosis based on non-invasive recordings. Our approach supports accurate and generalizable classification of dementia staging, combined with accurate estimation of future cognitive decline risk. The ability of DADD to reconstruct individual neurodegeneration levels provides deeper insights into disease progression, bridging the gap between network structure and cognitive outcomes. This method represents a scalable, generalizable and cost-effective solution for early AD detection, potentially facilitating widespread clinical implementation and improving patient management strategies.
References
[1] Amato, L. G., et al. (2024). Personalized modeling of Alzheimer’s disease progression estimates neurodegeneration severity from EEG recordings. Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring, 16(1), e12526. https://doi.org/10.1002/dad2.12526 [2] Amato, L. G., et al. (2025). Digital twins and non-invasive recordings enable early diagnosis of Alzheimer’s disease. Alzheimer’s Research & Therapy, 17(1), 125. https://doi.org/10.1186/s13195-025-01765-z [3] Amato, L. G., et al. (2026). Digital twins support cross-modal and cross-centric classification of mild cognitive impairment. Communications Medicine, 6(1), 30. https://doi.org/10.1038/s43856-025-01281-z
Acknowledgement This work was supported by the Italian Ministry of Research, in the context of the project NRRP “Fit4MedRob-Fit for Medical Robotics” Grant (# PNC0000007)
