Schedule of Events | Symposia

Novel Multi-Analysis of Binary Features in Ripple, Beta, and Gamma Bursts for SVM Classification of Memory-Related Event Order in iEEG

Poster Session B - Sunday, March 30, 2025, 8:00 – 10:00 am EDT, Back Bay Ballroom/Republic Ballroom

Hamed Aliyari¹ (hamed.aliyari@utsouthwestern.edu), Bradley Lega¹; ¹Department of Neurological Surgery, UT Southwestern Medical Center, Dallas, TX, U.S.A.

Remembering the order of events is a fundamental component of episodic memory. Cognitive theories propose that recall of a serial-ordered representation is driven by an internally maintained context that integrates new information over an extended time scale. This study obtained intracranial EEG (iEEG) recordings from 60 neurosurgical patients engaged in a serial recall task. The dataset encompassed simultaneous recordings from various brain regions, including the hippocampus, prefrontal cortex, and posterior cingulate cortex. A novel pattern extraction classifier, inspired by mechanisms of order memory observed in non-human primates and the temporal relationships among binary gamma bursts, beta bursts, and ripple events in working memory, was employed to investigate whether smoothed ripples, beta bursts, low gamma bursts, and high gamma bursts could serve as predictive features for recalling the sequence of events at each serial position. The analysis revealed that applying a modified Support Vector Machine (SVM) classifier to the orbitofrontal middle frontal gyrus (oMFG) in the right hemisphere, the orbital part of the inferior frontal gyrus (IFGorb) in the left hemisphere, and the middle temporal gyrus (MTG) in both hemispheres facilitated accurate identification of the correct word order in the sequence. These findings highlight the importance and applications that the innovative approach of smoothing neural features, such as ripples and bursts, can have for subsequent use in predictive modeling. Moreover, the classifier demonstrated its efficacy in addressing order memory tasks localized to the frontal cortex, thereby underscoring its potential to advance the understanding of neural mechanisms underlying episodic memory.

Topic Area: EXECUTIVE PROCESSES: Working memory