Predictive Modeling of Sleep Slow Oscillation Emergence on the electrode manifold: Toward Personalized Closed-Loop...

Area of Science:

• Neuroscience
• Computational Biology
• Sleep Science

Background:

• Sleep slow oscillations (SOs) are crucial for cognitive function and sleep's restorative properties.
• Enhancing SOs via brain stimulation offers potential clinical benefits, particularly for individuals with cognitive impairments.
• Current SO detection methods for closed-loop stimulation are computationally intensive and spatially limited.

Purpose of the Study:

• To develop a computational model for predicting SO occurrences across multiple electrode locations during sleep.
• To provide an alternative to real-time SO detection, enabling more flexible and personalized brain stimulation.
• To improve the accuracy of targeting SOs in brain stimulation applications.

Main Methods:

• Utilized polysomnography data from 22 subjects, including 64 EEG channels.
• Modeled SO occurrence in N3 or N2&N3 sleep stages using cumulative fitting with exponentials and a renewal point process.
• Employed an inverse Gaussian model to estimate SO timing probability density functions and parameters across sleep cycles.

Main Results:

• Observed a power-law decline in SO counts across sleep cycles, mirroring slow wave activity trends.
• SO timing models showed increasing trends in mean (μ) and shape (λ) parameters across cycles.
• Demonstrated that personalized SO prediction parameters can be derived from early sleep cycle data.

Conclusions:

• Established a predictive model for SO occurrence during NREM sleep, offering insights into their spatiotemporal organization.
• The model facilitates personalized stimulation paradigms by enabling prediction of SO timing.
• This approach enhances the precision of SO targeting in brain stimulation, advancing therapeutic potential.