Electrical brain stimulation and continuous behavioral state tracking in ambulatory humans

Area of Science:

• Neuroscience
• Biomedical Engineering
• Epilepsy Research

Background:

• Deep brain stimulation (DBS) treats drug-resistant epilepsy but can disrupt sleep.
• Intracranial electroencephalography (iEEG) offers continuous monitoring of brain activity.
• Assessing sleep quality is crucial for managing epilepsy and its comorbidities.

Purpose of the Study:

• To investigate the feasibility of automated sleep classification using iEEG during DBS in epilepsy patients.
• To evaluate the performance of a patient-specific sleep classifier under various DBS frequencies.
• To enable continuous, long-term sleep monitoring in ambulatory epilepsy patients.

Main Methods:

• iEEG data recorded from hippocampus (HPC) and anterior nucleus of thalamus (ANT) in four epilepsy patients.
• Development of a patient-specific sleep classifier using power band features from HPC iEEG.
• Validation against polysomnography (PSG) and prospective deployment during ANT DBS at different frequencies (0 Hz, 2 Hz, 7 Hz, >100 Hz).

Main Results:

• A patient-specific automated sleep staging classifier achieved an average F1-score of 0.894.
• The classifier accurately distinguished between awake, REM, and non-REM sleep across all tested DBS frequencies.
• Successful prospective deployment allowed for multi-month sleep pattern characterization in home environments.

Conclusions:

• Automated sleep classification using HPC iEEG is feasible during ANT DBS in epilepsy patients.
• This technology can reliably track behavioral states and sleep patterns.
• Continuous sleep monitoring can help optimize DBS therapy and manage associated comorbidities.