Clinical validation of an adapted Eleveld Model for high-dose propofol treatments for depression.

Repeated administration of high doses of propofol to patients with treatment-resistant depression (TRD) has been shown to produce antidepressant effects in small clinical trials. These effects can be elicited when the patient's EEG burst-suppression ratio (BSR) is maintained at 70-90% for 15 min in repeated treatments. This deep anesthesia domain lies beyond the range of current propofol pharmacokinetic/pharmacodynamic (PK/PD) models. In this study, we adapt the Eleveld model for use at deep anesthesia levels with a BSR endpoint, with the goal of aiding the estimation of the dosage of propofol needed to achieve 70-90% BSR for 15 min. We test the ability of the adapted model to predict BSR for these treatments. Twenty participants underwent 6-9 treatments of high doses of propofol (5-9 of which were included in this analysis) for a total of 115 treatments. To adapt the Eleveld model for this endpoint, we optimized the model parameters Ke0, γ and Ce50. These parameters were then used in the adapted model to estimate second-by-second BSR for each treatment. Estimated BSR was compared with observed BSR for each treatment of each participant. Median absolute performance error (MdAPE) between the estimated and observed BSR (25th-75th percentile) was 6.63 (3.79-12.96) % points and 8.51 (4.32-16.74) % between the estimated and observed treatment duration. This predictive performance is statistically significantly better at predicting BSR compared with the standard Eleveld model at deep anesthesia levels. Our adapted Eleveld model provides a useful tool to aid dosing propofol for high-dose anesthetic treatments for depression.

Analysis of single-unit firing patterns in multi-unit intrafascicular recordings.

A system for extracting single-unit activity patterns from multi-unit neural recordings was tested using real and simulated neural data. The system provided reliable estimates of firing frequency for individual units in simulated multi-unit data and allowed reliable determinations of the responses of individual cutaneous mechanoreceptor units to 'natural' stimuli such as brushing or pressing on the skin. An implementation of the system, which operated online and in real time, was used to obtain estimates of multiple, single-unit responses from multi-unit intrafascicular electrode recordings. The pattern of activity across the population of units in a given recording gave a reliable indication of the type of stimulus that had evoked the activity. It was concluded that this system, used in combination with intrafascicular peripheral nerve recordings, could be used to provide online, real-time information about peripheral stimuli.