Automated Sleep Classification Using iEEG from Hippocampus and Anterior Nucleus of Thalamus
Abstract number :
2.099
Submission category :
3. Neurophysiology / 3G. Computational Analysis & Modeling of EEG
Year :
2023
Submission ID :
469
Source :
www.aesnet.org
Presentation date :
12/3/2023 12:00:00 AM
Published date :
Authors :
Presenting Author: Filip Mivalt, MS et MS – Mayo Clinic
Vaclav Kremen, PhD – Mayo Clinic; Vladimir Sladky, MS – Mayo Clinic; Nicholas Gregg, Dr – Mayo Clinic; Kai Miller, Dr – Mayo Clinic; Jamie van Gompel, Dr – Mayo Clinic; Benjamin Brinkmann, Dr – Mayo Clinic; Erik St Louis, Dr – Mayo Clinic; Gregory Worrell, Dr – Mayo Clinic
Rationale:
High frequency (HF) electrical brain stimulation (EBS) of the Anterior Nucleus of the Thalamus (ANT) is an established therapy for drug resistant focal epilepsy. However, sleep disturbances are a common comorbidity of epilepsy and it has been shown that HF ANT EBS may further exacerbate sleep disruption. Hippocampal (HPC) iEEG data recorded using a novel implantable device was previously used for reliable automated sleep scoring. Here, we investigate the feasibility of automated sleep classification using ANT iEEG that can benefit a standard clinical device with SANTE stimulation.
Methods:
Three people with epilepsy and implanted with the investigational Medtronic Summit RC+STM underwent simultaneous polysomnography (PSG) and iEEG monitoring for up to three nights. Sleep hypnograms were created using AASM2012 rules and were utilized to develop automated sleep classifiers for Awake, rapid-eye-movement (REM), and non-REM sleep stages.
Results:
An automated sleep iEEG classifier using a single channel iEEG was designed trained and validated for HPC and ANT. The achieved overall F1-score performance was 0.785 for HPC and 0.830 for ANT data. The models were deployed to a long-term iEEG data spanning over 3 years in total to create sleep-wake profile.
Conclusions:
We have demonstrated the feasibility of automated sleep classification using ANT iEEG which performs better than automated HPC iEEG sleep classifiers. In the future, ANT iEEG sleep classifiers may enable optimization of EBS therapy based on behavioral state profiling and help to mitigate sleep disruptions caused by EBS.
Funding:
This research was supported by National Institutes of Health (UH2&3-NS95495, R01-NS092882), LQ1605 from the National Program of Sustainability II (MEYS CR, Czech Republic), and institutional resources from Mayo Clinic, Rochester MN USA, Medtronic Plc, Minneapolis, MN, USA.
Neurophysiology