Objective. Reliable closed loop infusion systems for regulating paralysis l
evel can be a great convenience to the anesthesiologists in automating thei
r task. This paper describes the in vivo performance evaluation of a self-t
uning controller that is designed to accommodate large variations in patien
t drug sensitivity, drug action delays and environmental interfering noise.
Methods. The infusion system was evaluated in six adult mongrel dogs. Foll
owing the manual induction of paralysis by an anesthesiologist, the control
ler regulated the infusion of vecuronium to maintain a desired level of par
alysis. The integrated EMG response of the hypothenar muscle to a train-of-
four stimulation of the ulnar nerve quantified the depth of paralysis. The
controller's robustness was tested by contaminating the sensed twitch signa
l with electrocautery noise and electrode disconnection. Results. The contr
oller reached the initial level of paralysis of 100% in about 4.0 minutes a
nd arrived at the desired level of 90% with an overshoot of 6.38% (+/-6.82)
. it maintained the desired level of paralysis with a 2.04% (+/-1.20) mean
offset at 90% and 0.4% (+/-0.5) mean offset at 80% steady state level, resp
ectively. The mean infusion rate to sustain 90% and 80% paralysis were 2.70
(+/-2.05) and 2.15 (+/-2.57) ((mg/kp)/min), respectively. Conclusions. The
system adapted to a large variation in the sample subject drug sensitivity
It remained stable despite large amplitude disturbances and maintained the
paralysis at the desired level following the removal of the disturbances.