ALZHEIMER’S AND PARKINSON’S: TWO FACES OF ONE DISEASE. MRI AND EXPLAINABLE MACHINE LEARNING VALIDATION OF THE NES COMMON ROOT HYPOTHESIS

Increasing evidence suggests that Alzheimer’s (AD) and Parkinson’s disease (PD) are not separate entities but divergent manifestations of a shared neurodegenerative root. The Neurodegenerative Elderly Syndrome (NES) hypothesis frames this process through three stages: seeding (initial monoaminergic/protein dysfunction), compensation (homeostatic symptom masking), and bifurcation (phenotypic divergence). This study validates the bifurcation stage using longitudinal structural MRI data integrated over 24 months (0, 12, 24 months) to represent the brain state during early clinical manifestation. Multiple explainable machine learning (XML) architectures decode structural signatures across 15 key brain regions. Findings reveal a striking divergence in homeostatic resilience: while AD exhibits a catastrophic structural collapse (p < 0.0001) with 95.1% classification accuracy, PD maintains a "structural silence" nearly indistinguishable from healthy controls (75.6% accuracy, p > 0.05). This gap, coupled with brainstem volumetric stability in PD, provides quantitative evidence of a robust monoaminergic compensation—mediated by noradrenergic and serotonergic nuclei—masking macroscopic decay despite active seeding. Conversely, hippocampal and limbic collapse in AD marks the terminal failure of these homeostatic systems post-bifurcation. Furthermore, gender-specific analysis reveals significant amygdala volume loss in female PD patients, suggesting sex-dependent susceptibility trajectories. In conclusion, AD and PD represent opposite endpoints of a unified neurodegenerative continuum. Their structural differences reflect the differential velocity of compensatory failure rather than separate etiologies. These results validate MRI as a NES biomarker, advocating for unified therapeutic interventions targeting early homeostatic preservation.