Recent developments in Process Systems Engineering as applied to medicine

https://doi.org/10.1016/j.coche.2012.09.007Get rights and content

The paper contains a review of recent work by Process Systems Engineering (PSE) researchers in biomedical systems. The review is organised in the following four categories: analysis of systems corresponding to diagnosis of a condition, model simulation providing predictions resulting in a potential use for prognosis, design for an optimal action or set of actions based on a model and set of objectives which corresponds to devising a therapy, and operations where sets of patient data may be used to manage patient well-being or a disease condition. There has been work particularly in glucose regulation involving the liver system (the body's ‘chemical factory’), in the brain, in the pulmonary system, and to a lesser extent in some other parts of the body. There is evidence of useful proposed clinical outcomes. So far there is little evidence of clinical use of the results. There is a need for better models, handling of uncertainty to ensure conservative (and where possible guaranteed) predictions, and for greater interaction with the clinical community. There is real potential for much greater use of PSE approaches to these complex systems.

Highlights

► We review Process Systems Engineering methodologies applied to physiological systems. ► There are results for diagnosis, prognosis, and design of therapies. ► Main results involve glucose regulation and blood flow in the brain. ► There is no evidence so far for clinical use. ► Greater interaction with clinicians is required to gain experience and confidence.

Introduction

As Chemical Engineering has spread its wings well beyond manufacturing so too have some Process Systems Engineering researchers. An area of relatively recent interest has been in biomedical systems and this paper will review recent work in this area by Process System Engineering researchers. The domain of Process Systems Engineering is the analysis and design of complex systems involving chemical and physical changes. The human body is a very complex system involving many chemical, electrical and mechanical actions. It consists of many complex interacting subsystems which together maintain the functioning of the body.

Medical practitioners have benefited from Engineering advances most notably in imaging advances and prosthetic devices. In this review I will concentrate on developments that involve decisions based on models of physiological phenomena. I will of course refer to some developments at the cellular level but only where they affect physiology directly.

Some authors have noted the potential that Process Systems Engineering has for biomedical systems: through modelling, experiment, and systems integration [1] and through integrating imaging, modelling and Computational Fluid Dynamics [2, 3••]. There has been considerable work done on the brain, particularly in tandem with advanced imaging capabilities, on the pulmonary system, and with the formation of metastatic tumours. The liver acts as the body's chemical factory with a central role in maintaining blood glucose levels and in filtering out waste products. As such it is not surprising that there is a body of work that addresses aspects of modelling and control of the liver system [4] and this work will form the mainstay of the discussion in this review. Medical practitioners confront many of the problems that we are familiar with in complex systems: dealing with measurement uncertainty, lack of understanding which leads to model errors, multiple solutions and optima, and having many variables.

This review was undertaken by looking at the recent Process Systems Engineering literature. I have considered model-based approaches to the following problems as understood by Process Systems Engineers: Analysis, Simulation, Design and Operations. In analysis we seek to better understand the phenomena that are causing a particular observed outcome. In medicine this corresponds to making a diagnosis of a patient. We use simulation to predict potential outcomes of a set of scenarios which in medicine is best seen as making a prognosis. As engineers we design systems to have a particular purpose or to improve an existing system and this in medical systems corresponds to devising a therapy which can alleviate or remove a problem. Finally we use PSE tools to devise optimal operation strategies which corresponds to devising systems which maintain well-being or for managing a disease to avoid deterioration. There has been work in all of these areas except the last. The paper is organized according to this taxonomy. I have chosen only to include contributions where there is some chemical change involved.

Section snippets

Analysis  diagnosis

The value of modelling in Systems Biology has often been said to be in helping to understand the consequences of interacting complex phenomena. In this section I will highlight some examples of where Process Systems Engineering and researchers in this community have contributed to approaches to building understanding of the phenomena and causes of measurable symptoms. Once a model has been constructed its richness can be explored by comparison with data, either from experiments or from

Simulation  prognosis

Once validated a model can be used to predict potential outcomes on the basis of alternative assumptions or of possible external stimuli. The medical profession makes predictions about potential outcomes based on their understanding of the causes of the symptoms (measurements) that they are aware of. With properly validated models there is a role for generating output simulations to aid the medical profession to follow the consequences of multiple phenomena and their synergistic effects. This

Design  therapy

The ultimate value of models is to be able to devise therapeutic ‘designs’, that is, courses of action which will alleviate or improve a condition. The human system is so complex that to be able to foresee all potential outcomes of a clinical course of action is impossible even if the diagnosis is correct. A model could help with this although of course as with any model-based decision support system users must be very careful not to use a model beyond its limits of validity. I have included in

Operations  well-being and disease control

The final area I will consider briefly is the use of data driven approaches to ‘operations’ of human systems: the maintenance of the well-being of individuals and the medium and long-term control of chronic conditions.

A lot of patient data are available for specific conditions that have been built up over many years which could be used to generate a general picture of specific conditions and consequences of therapies. Such data can be from routine measurements in clinic or at home, from

Conclusions

The review demonstrates the growing interest by our community in applying Process Systems Engineering techniques to biomedical problems. The tools seem adequate in the range of analysis/diagnosis and simulation/prognosis problems that have been attempted although there is little yet that ties extensive patient data to the models. The design of therapeutic interventions is almost exclusively limited to controlling drug delivery on the basis of very focused models without links to models of wider

References and recommended reading

Papers of particular interest, published within the period of review, have been highlighted as:

  • • of special interest

  • •• of outstanding interest

References (40)

Cited by (0)

View full text