High Frequency Oscillations (HFOs) is an important biomarker that can potentially pinpoint the epileptogenic zones (EZs). However, the duration of HFO is short with around 4 cycles, which might be hard to recognize when embedded within signals of lower frequency oscillatory background. In addition, annotating HFOs manually can be time-consuming given long-time recordings and up to hundreds of intracranial electrodes. We propose to leverage a Switching State Space Model (SSSM) to identify the HFOs events automatically and instantaneously without relying on extracting features from sliding windows. The effectiveness of the SSSM for HFOs detection is fully validated in the intracranial EEG recording from human subjects undergoing the presurgical evaluations and showed improved accuracy when capturing the HFOs occurrence and their duration.