Epilepsy affects 1% of the population, with up to one-third of patients being medication-resistant. Surgery is the only curative treatment, yet over one-third of surgical patients fail to achieve seizure freedom due to the lack of a reliable epileptogenic zone (EZ) biomarker. We introduced and validated mini-seizures, frequent hypersynchronization events at EZ hubs that mirror seizure network dynamics, as a novel interictal EEG biomarker. Using a dynamical networks-based model, we analyzed short interictal intracranial EEG from 159 patients across two institutions. Our model, integrating hypersynchronous network properties and clinical data, successfully identified EZ hubs and accurately predicted one-year postoperative seizure outcomes. Our model (mean F1 score: 87%) outperformed the high-frequency oscillations-based model (mean F1 score: 79%) and seizure onset zone resection-status-based model (current clinical standard) (mean F1 score: 78%), supporting its potential as a robust interictal biomarker for EZ localization. Our findings suggest mini-seizures and seizures exist on a continuum of epileptic events, sharing common network properties. Unlike seizure-based analyses that require 1-2 weeks of monitoring to capture spontaneous seizures, mini-seizures provide a rapid alternative using only brief interictal recordings.