A Dynamical Digital Twin Unmasks Hidden Neuromotor Control Policies and Catastrophic Tipping Points in Parkinson's Disease

Abstract

How the central nervous system maintains upright stance--and why this capability collapses in Parkinsons disease (PD)--remains a fundamental open question in computational neuroscience. A primary barrier is a severe topological degeneracy between observable sway kinematics (z-space) and latent neural control policies (-space), causing radically distinct control strategies to mask as identical sway patterns. Here, we establish a dynamical digital twin framework bridging empirical sway time series with mechanistic physical models. Assimilating a large cohort (N = 1, 038) into an intermittent control model via Bayesian inference, we construct a bidirectional mapping (z {leftrightarrow} ) that unmasks these hidden dynamics. We prove that healthy stance is universally governed by flexible intermittent control near an optimal intermittency ratio ({rho} {approx} 0.5), whereas PD progression reflects a structural regression toward rigid continuous control ({rho} [->] 1). Crucially, parameter topology in -space reveals that neural control policies reside on a folded low-dimensional manifold separated by a distinct parameter gap; attractor bifurcation analysis demonstrates that a catastrophic tipping point is embedded within this void, proving that parkinsonian postural breakdown is driven by a discontinuous dynamical phase transition rather than continuous control decay. By unmasking hidden disease severity beneath degenerate sway phenotypes, our framework reframes parkinsonian motor failure as an attractor bifurcation on a neural control manifold, providing a modern computational realization of the classical "dynamical disease" paradigm. SignificancePostural collapse in Parkinsons disease presents a profound neurobiological paradox: cellular degeneration progresses continuously over decades, yet clinical motor failure strikes as a sudden, catastrophic collapse. We resolve this paradox by demonstrating that motor breakdown is not a passive decay of control resources, but a discontinuous dynamical phase transition (attractor bifurcation) triggered when underlying neural control policies cross a latent parameter gap. By unmasking "phenotypic degeneracy"--where radically different neural control strategies produce identical body sway--our digital twin framework redefines neurodegenerative motor failure as a qualitative regime shift in complex control dynamics, offering a universal paradigm for predicting catastrophic transitions in neurological disorders.