knowledge transfer


Smart Adaptive Control of Muscle Relaxation

Submitted by D A Linkens with S B Hasnain,29.11.2001,IBA D


To provide feedback control of neuromuscular blockade during surgical operations using continuous drug infusion. In this case study the challenge is to provide an adaptive controller which will accommodate a change in the type of drug being infused, without being re-programmed. Measurement signals are available from a Relaxograph (or equivalent monitor) which uses supra-maximal stimulation of the ulnar nerve at the wrist and processed signals from the palm of the hand.


The feasibility of automatic control of muscle relaxation had already been established using classical PI algorithms. This had been done using simulation and in clinical trials. This has been shown that there can be 4:1 variability in patient gain sensitivity to a particular drug. Hence, the Self-Organising Fuzzy Logic Control (SOFLC) structure of Procyk and Mamdani has been adopted for this application, so that the controller can adjust its rules to be patient-specific. Basically, the SOFLC is similar to a PI controller with fuzzy rules being elicited via on-line self-exploration. The results of this are described in an accompanying case study.

The SOFLC structure has also been studied for the case of porting the system from the use of one type of drug to another one. In this case, the SOFLC is initiated to suit the pharmacokinetics of one drug (established either via simulation or clinical trials) and then allowed to adapt on-line to the pharmacokinetics of the new drug.

Status and results

The SOFLC system has been shown to give good control performance ( in terms of stability and settling time)under simulated conditions when switching between infusion of pancuronium (a slow-acting drug) to atracurium (a fast-acting drug).Successful results were also obtained if the SOFLC was initialised with zero rules (i.e. no a priori knowledge of drug dynamics),but more cautious control excursions were necessary to ensure stability for all conditions. This can be achieved by adjustment of the scaling factors which are part of the SOFLC architecture.

Adaptivity and portability

The SOFLC can achieve Level 2 adaptation in the EUNITE definition, in that it can adjust its control to new drug usage. This is via prior initialisation to an alternative drug. It approaches Level 3 adaptivity, in that it can commence from a zero rule base. However, this requires slower adaptation than for Level 2.Also,in this architecture some basic knowledge of the overall system gain and time constants is needed for initial settings of the scaling factors within the SOFLC.

More information

The work was done as part of a PhD project on the use of parallel computing using self-organising structures. Details are in:

S B Hasnain,"Self-organising control systems and their transputer implementation", PhD thesis, University of Sheffield,UK,1989.

technical reports & case studies

executive summaries

scientific papers




search eunite with