Work

Spurious Resilience and Accelerated Localized Collapse: A Topological Phase Transition Analysis of Alzheimer's Disease Structural Connectomes Based on Real DTI Data and Dual-Seed Validation

Year: 2026
Type: preprint
Abstract: The prevailing consensus posits that the brain network in Alzheimer's disease (AD) becomes increasingly vulnerable with pathological accumulation, ultimately leading to global topological collapse. However, based on targeted topological attacks on real diffusion tensor imaging (DTI) structural connectomes from 76 ADNI patients and 49 healthy ageing subjects, we reveal a counterintuitive physical phenomenon: AD networks exhibit significant "spurious resilience" at the macroscopic percolation phase transition threshold. Under a strictly fixed network density of 15%, the percolation threshold of the AD group (≈76.5%) paradoxically exceeded that of the healthy group (≈73.1%). To ensure the absolute robustness of our findings, we introduced an independent dual-seed reproducibility validation (Seed=42 and Seed=100), yielding an error margin of <0.3%. We propose that this "spurious resilience" arises from pathologically elevated local clustering coefficients in AD, forming numerous "hyper-connected local cliques" or "topological dead-knots." These dead-knots act as topological buffers, maintaining the apparent integrity of the Giant Connected Component (GCC) even after global efficiency has already collapsed. This discovery not only explains why global structural compensation masks early network degeneration but also delineates a new paradigm for clinical interventions (e.g., rTMS), emphasizing that "breaking local dead-knots is superior to global repair."
Language: English
Open Access status: green
OpenAlex ID: W7233606140
Cites:
Cited by:
Related to: