Reduced Persistence of Cortical States in Benign Paroxysmal Positional Vertigo: A Hidden Markov Model Analysis of Resting-State Functional Magnetic Resonance Imaging

Abstract To determine whether hidden Markov model (HMM) analysis of resting-state functional MRI (fMRI) can identify altered temporal persistence of recurring cortical states in patients with benign paroxysmal positional vertigo (BPPV). This cross-sectional study included 70 patients with BPPV and 46 healthy controls. Time series from 100 cortical parcels were analyzed using a five-state Gaussian HMM specified a priori. Switching rate was the primary outcome, and mean empirical self-transition probability was a supporting measure. Group comparisons were adjusted for age, sex, years of education, and mean framewise displacement, with false-discovery-rate correction. Sensitivity analyses evaluated head motion, model orders K = 2–15, and omission of global signal regression. Switching rate was higher in patients with BPPV than in controls (0.417 ± 0.030 vs 0.370 ± 0.035; adjusted β = 0.047; 95% confidence interval, 0.035–0.060; q = 3.61 × 10 − 11). Mean empirical self-transition probability was lower in the patient group (0.573 ± 0.027 vs 0.623 ± 0.035; adjusted β = −0.049; 95% confidence interval, − 0.061 to − 0.037; q = 8.04 × 10 − 12). Patients with BPPV also showed higher fractional occupancy of State 2 and lower fractional occupancy of State 3. Both global effects remained directionally consistent across sensitivity analyses. BPPV was associated with reduced persistence of recurring cortical states on resting-state fMRI. Temporally resolved HMM analysis may complement conventional time-averaged functional imaging measures.