An adaptive brain-machine interface algorithm for control of burst suppression in medical coma

Area of Science:

• Neuroscience
• Biomedical Engineering
• Control Systems

Background:

• Burst suppression, an EEG pattern, indicates brain inactivation and is targeted in medical coma.
• Automating medical coma control via a brain-machine interface (BMI) requires adaptive models for drug dynamics and EEG observations.
• Current methods necessitate offline parameter fitting, limiting real-time adaptation to changing conditions.

Purpose of the Study:

• To develop a novel adaptive algorithm for robust control of medical coma.
• To enable real-time estimation of drug concentrations and system parameters.
• To achieve precise control of time-varying burst suppression trajectories with minimized infusion rate variations.

Main Methods:

• Developed an adaptive recursive Bayesian estimator for joint real-time estimation.
• Designed a linear-quadratic-regulator controller incorporating estimates for robust feedback control.
• Implemented simulations to validate the adaptive algorithm's performance.

Main Results:

• The adaptive algorithm achieved precise control of time-varying burst suppression levels.
• Effective control was demonstrated even with randomly initialized model parameters.
• The algorithm significantly reduced infusion rate variations during steady-state conditions.

Conclusions:

• The proposed adaptive algorithm offers robust and precise control of medical coma.
• Real-time parameter estimation enhances the adaptability of BMI-controlled drug delivery.
• This approach improves patient management by enabling stable and responsive coma induction.