Model-based analysis and forecast of sleep-wake regulatory dynamics: Tools and applications to data

Area of Science:

• Neuroscience
• Computational Biology
• Systems Biology

Background:

• Sleep-wake regulation and state of vigilance (SOV) are intrinsically linked to neurological disorders like schizophrenia and epilepsy.
• Understanding the bidirectional relationship between disease severity and sleep disturbances requires investigating neurophysiological interactions within the sleep-wake regulatory system (SWRS).

Purpose of the Study:

• To develop and validate a data assimilation (DA) framework for reconstructing and predicting SWRS states in normal and pathological brains.
• To estimate time-dependent model parameters crucial for accurate system state reconstruction and prediction.

Main Methods:

• Utilized unscented Kalman filter-based DA with physiologically based mathematical models of the SWRS.
• Developed a novel estimation method for simultaneous fitting and tracking of multiple model parameters.
• Incorporated fixed-lag smoothing to enhance reconstruction of system inputs and delayed effects.
• Experimentally recorded brain activity from freely behaving rodents and classified discrete SOV.

Main Results:

• Successfully reconstructed and predicted SWRS states using experimental rodent brain activity data.
• The DA framework accurately estimated time-dependent model parameters from discretized SOV observations.
• Demonstrated the ability to forecast future SWRS states and transitions from out-of-sample recordings.

Conclusions:

• The developed DA framework provides a robust method for analyzing and predicting SWRS dynamics.
• This approach facilitates a deeper understanding of the neurophysiological underpinnings of sleep disturbances in neurological disorders.
• The framework holds potential for advancing research into brain function and disease mechanisms.