Generalized sleep decoding with basal ganglia signals in multiple movement disorders.

Area of Science:

• Neurology
• Biomedical Engineering
• Sleep Medicine

Background:

• Sleep disturbances significantly impair quality of life in neurological disorders.
• Closed-loop deep brain stimulation (DBS) requires automated sleep stage decoding from intracranial signals.
• Current methods lack generalization across diverse patient populations and neurological conditions.

Purpose of the Study:

• To develop a generalized, multi-class, sleep-specific decoder for intracranial signals.
• To improve automated sleep stage decoding for closed-loop DBS applications.
• To enhance sleep disturbance management in movement disorders.

Main Methods:

• Leveraged overnight polysomnography and basal ganglia local field potentials from 121 patients with various movement disorders.
• Developed and validated a generalized sleep decoder named BGOOSE.
• Investigated the impact of electrocorticography on decoding performance and proposed an optimal decoding map.

Main Results:

• The BGOOSE model achieved 85% average accuracy across patients and disease conditions.
• The decoder demonstrated robustness with recordings from different basal ganglia targets.
• An optimal decoding map facilitated channel selection for improved model performance.

Conclusions:

• BGOOSE is a powerful tool for generalized sleep decoding from intracranial signals.
• This technology holds potential for precision stimulation delivery in DBS.
• BGOOSE can lead to better management of sleep disturbances in movement disorders.