Current clinical practice of anesthesiology involves administering anesthetics using standard dosing guidelines, monitoring physiological responses during administration, and adjusting the dose depending on the patient responses. Importantly, direct monitoring of the brain is not part of standard anesthesiology practice, despite the desired effect taking place in the brain. In cases where the brain is directly monitored, proprietary, drug-independent, non-personalized depth of unconsciousness indices are commonly used to infer conscious state from electroencephalography (EEG) signals. This may result in overdosing, as there are significant inter-individual differences in neurological response to identical doses of anesthetics and signatures of anesthesia vary significantly with age or with choice of anesthetic. Consequently, older patients are often given more anesthetic than necessary and are at high risk for developing post-operative cognitive dysfunction (POCD) and delirium, which may last up to several months. The prevalence of surgical procedures on older patients will rise as the population ages, necessitating new approaches for ensuring safe, personalized delivery of anesthetics that ensure unconsciousness, but also safety and return to normal cognitive function following surgery. This project seeks to develop closed-loop control of general anesthesia in humans as a means of personalizing anesthesia care.
Aim 1 of this project seeks to use machine learning to develop a drug- specific marker of depth of unconsciousness that reflects the varying signatures of unconsciousness with age and may be understood clinically in terms of brain function.
Aim 2 of this project is to develop models and nonlinear model predictive control (MPC) algorithms for regulating depth of unconsciousness during general anesthesia. MPC is a control scheme that may be easily modified to incorporate clinical safety features and has been used extensively in medical control systems. To tightly regulate depth of unconsciousness, fast actuation by the anesthetic is critical.
Aim 3 of this project seeks to characterize the effect of propanidid, a fast-acting anesthetic currently used in clinical practice in Mexico. The culmination of Aim 2 will result in designing clinical trials for the first closed- loop anesthetic control and the culmination of Aim 3 will result in designing a clinical pharmacokinetic study of propanidid. The results of this project have the potential to transform clinical practice of anesthesia and allow individualized anesthesia care. By direct EEG monitoring and control of brain state rather than secondary markers, the personalized treatment developed by this project should help reduce overdosing of senior surgical patients and reduce incidence of POCD and delirium.
The proposed work will develop algorithms for monitoring and regulating depth of unconsciousness during general anesthesia, with an emphasis on understanding how neural signatures of unconsciousness change as the brain ages. This project will also characterize the effects of propanidid, a fast-acting anesthetic ideal for use in a control system. The results of this research will improve our ability to tightly regulate depth of unconsciousness through automation and reduce the incidence of anesthetic overdosing and postoperative delirium through personalized medicine.
