Automatic Closed-Loop Administration of a Vasoactive Drug
Nicolo Malagutti (ANU)
APPLIED SIGNAL PROCESSING SERIESDATE: 2011-08-11
TIME: 11:00:00 - 12:00:00
LOCATION: RSISE Seminar Room, ground floor, building 115, cnr. North and Daley Roads, ANU
CONTACT: JavaScript must be enabled to display this email address.
ABSTRACT:
Patients undergoing major surgery often require administration of vasodepressant drugs to suppress intraprocedural arterial pressure peaks caused by painful stimuli and to control postoperative hypertension. Research into viable strategies to design and implement an automatic closed-loop system capable of administering vasodepressants and adjusting the dose on the basis of the patientas measured blood pressure has been ongoing for the past two decades. None of the solutions proposed thus far, however, have been embraced in the real clinical setting. Still, the potential advantages of such a solution remain unquestioned: a successful automatic system would reduce the risk of under/overdosing and therefore improve health outcomes; also, it could reduce healthcare costs by replacing human operators in the carrying out of the time-consuming task of periodic dose adjustment. We suggest that the failures of previous approaches may be a consequence of overly complex designs, which may discourage routine use by clinicians, or the existence of simplifying assumptions which may be violated in practice, giving rise to unresolved risk concerns.
Critical problems in this application include the time delay associated with the transport of the drug in the bloodstream, the considerable variability in the response to the drug among patients and the fact that the response can even vary for one patient over time. Our proposal for a new control architecture draws on recent control engineering results in the design of robust multiple model adaptive control (MMAC) systems. MMAC is an architecture in which the behaviour of the plant (patient) is matched with that of one of a finite number of candidate mathematical models. On the basis of this correspondence, a specific controller designed for the best matching model is used in the feedback loop where it is expected to yield satisfactory performance. Provided that a suitably large number of models is used, MMAC can effectively deal with non-linear and even time-varying processes by interpolating several locally valid linear models. This is an advantage for engineers as a wealth of control analysis and design techniques for linear systems can be utilised to ensure robust stability and performance.
Our system involves five candidate models and five corresponding controllers designed using a μ-synthesis method. It also utilises an algorithm based on Kalman filtering to identify the best matching model out of the candidate set. Initial simulation studies have shown promising results in the control of mean arterial pressure in the highly-uncertain, poorly-observable and time-varying patient system, even in high-noise settings. The system is currently undergoing further refinements to ensure maximum safety, which will be required before clinical implementation can be contemplated. An attractive aspect of this approach is that it could be easily modified and adapted to other instances of drug administration in which similar issues of variability in dose response exist.
BIO:
Nicolo Malagutti holds a Bachelor degree of Biomedical Engineering and a Masters degree in Industrial Engineering with a Biomedical Engineering major, both from Politecnico di Milano, Milan, Italy. He has worked as a Research and Development engineer with biomedical companies in the past. He is currently researching towards his PhD at RSISE with a project concerning automatic closed-loop drug administration.
