Abstract: Natural disasters disrupt local economies unevenly, but most post-disaster analyses use aggregated indicators that mask establishment-level dynamics in rural regions. This study develops a data-driven mobility modeling framework to quantify post-Hurricane Helene recovery using point-of-interest (POI) visitation time series. We analyze 5,671 establishments in rural western North Carolina using 616 days of mobility-derived visits (May 2023–Feb 2025), integrated with precipitation, elevation, and road-closure records. Counterfactual visitation trajectories are estimated with four forecasting models—SARIMAX, Prophet, Chronos-T5, and TimesFM—showing transformer-based methods outperform classical approaches, with TimesFM providing the most accurate and stable long-horizon forecasts. Disruption is measured via a normalized signed Area Under the Curve (AUC) between predicted and observed visits. Results indicate sectoral differences dominate geographic variation: education and public administration exhibit the largest declines, while retail and accommodation recover faster. Sector type, baseline visitation, and rainfall intensity are the strongest predictors of disruption.