Saturday, 26 July 2014

Michael Jamieson and the Yerkes-Dodson rollercoaster

Michael Jamieson is a Scottish swimmer who was tipped for gold (by Adrian Moorhouse and Rebecca Adlington) in the 200m breaststroke at the 20th Commonwealth Games in Glasgow. Jamieson also declared that he was aiming for a new world record. On the day, however, he was out-swum by fellow Scot, Ross Murdoch. Could the high expectations and amount of stress Jamieson was under have prevented him from achieving his full potential?


In 1908 Robert Yerkes and John Dodson wrote a paper entitled: "The relation of strength of stimulus to rapidity of habit formation". Using 40 "dancer" mice, a choice of two chambers and electric shocks of varying intensity, Yerkes and Dodson made an interesting discovery. As they increased the shock intensity the mice would learn faster which chamber to avoid. But only up to a point. Past this point the increase in shock intensity had a detrimental effect on the retention of this information.

The Yerkes-Dodson theory has been applied to human performance under stress. This means that people have an optimum point of stress and retention of learning. The optimum point is the plateau of the curve but each person will have different curves based on their personalities, experience and expertise.

The Yerkes-Dodson curve has also been shown to apply when we compare stress (or arousal) with performance (rather than habit-formation). Once the optimum point has been reached, then the greater the stress, the poorer the performance

The alternative route

The "original" U-shapes (top 2)
and the "easy task" bottom line
There is a slight addition to the Yerkes-Donaldson rollercoaster, the little-discussed, alternative sigmoidal route, which is based on the difficulty of the task. If the chambers were designed so that it would be difficult for the mice to distinguish between them, then the above findings held true. However, if it was very easy to distinguish between the two chambers then the higher the shock intensity, the faster the habit formation. This is shown in the graph as the bottom-most line. This means that, for simple tasks, the more stressed/aroused you are the faster you will learn/perform.

Relevance to simulation

If the Yerkes-Dodson theory is true then the amount of stress we expose our participants to will affect their learning and performance. Either too little or too much stress will have a negative impact. Likewise, as the task becomes more complex, the higher the likelihood that we will push the participant onto the downward slope. We must therefore design our courses with this knowledge in mind.
In addition, we must appreciate that inter-professional courses and courses which simultaneously feature both junior and senior staff are more likely to have a wider spread of curves. This means that the scenarios must be designed to stress the different healthcare personnel appropriately and not aim for the lowest common denominator.

Lastly, the Yerkes-Dodson theory can be applied to your facilitators as well. Too little challenge and they'll fall asleep, too much and they'll perform poorly. We need to make sure that we match the facilitators to the participants and have back-up available if needed.

Wednesday, 16 July 2014

Book of the Month: Modeling and Simulation in Biomedical Engineering by Willem van Meurs

About the author

Willem van Meurs received his doctorate in control engineering from Paul Sabatier University, Toulouse, France in 1991. His claim to fame in the field of simulation is that he is the co-inventor of the Human Patient Simulator (HPS) which was commercialised by Medical Education Technologies, Inc. (METI), now part of CAE Healthcare. In addition he was president of the Society in Europe for Simulation Applied to Medicine (SESAM) from 2005 to 2007.

Who should read this book?

According to van Meurs, the target audience is "those studying or working in biomedical engineering: engineers, physicists, applied mathematicians, but also biologists, physiologists and, clinicians", as well as "clinical educators and simulator technicians" using the HPS.
In reality, the number of people world-wide who would read the entire book (other than book reviewers) is probably in the hundreds.

I haven't got time to read 185 pages…

Thankfully, unless you are involved in designing or building an actual simulator, you can get away with only reading the following chapters.
  • Chapter 1: Introduction (10 pages providing an overview of the concepts and vocabulary)
  • Chapter 2: Model Requirements (9 pages discussing the initial stages of simulator design)
  • Chapter 12: Design of Model-Driven Acute Care Simulators (8 pages discussing training needs, training programme design and, as a result, simulator design)

What's good about this book

The book makes one appreciate the complexity behind the HPS, which aims to realistically model a human being's physiology. Even if we consider just one variable, for example the partial pressure of oxygen in arterial blood (PaO2), the HPS tries to consider a number of inputs (the partial pressure of oxygen in the inhaled gas, the degree of shunt in the lungs, the metabolic rate, the concentration of haemoglobin, etc.) and a number of outputs (respiratory rate, heart rate, cardiac contractility, ECG morphology, etc.). A number of these inputs depend on other variables and the outputs have a number of effects, some of them in a positive or negative feedback loop on the original variable, PaO2.

The book also introduces some useful terms in model development such as the concept of black, grey and white boxes. The familiar black box is used in modelling where there is no need to know what the internal workings of the model are: a given input results in a given output. A white box is where one needs to know the exact mechanics of how input becomes output, usually because these mechanics are influenced by other processes and therefore the output will not just depend on the given input but also on the state of the system. Lastly, a grey box is where some of the mechanics are known (and modelled) and others are not.

van Meurs also covers the development of a full-scale simulator, which has 4 steps:

  1. Conceptual model (with consideration of the qualitative aspects)
  2. Mathematical model (with consideration of the quantitative aspects)
  3. Software implementation (with consideration of interfacing)
  4. Simulation results and validation (with consideration of the output data)

Lastly, for the biomechanical engineer, there is a review section at the end of some of the chapters which ask you to use some of the material covered to work through problems. (Somewhat criminally however, van Meurs does not provide the answers (or at least a guide to obtaining the correct answer) in the book.)

What's bad about this book

And so it is proven...
In terms of the target audience, this must have been a difficult book to write. Trying to make the material easy enough for clinicians to grasp but in-depth enough for biomechanical engineers, van Meurs probably satisfies neither. It may have been better to write two books for the two audiences. Out of the two audiences, the biomechanical engineers probably got the better deal from this book. They will have no problem with the equations such as that shown here, which are liberally sprinkled throughout and they are also given a reasonable introduction to the basic variables in cardiorespiratory physiology.

There's a forgivable typo on p.58 where we're told that 1 hour = 1 hour = 3600 seconds. On p. 148 and 149 there is some missing information around chest wall and lung compliance. We are told that lung compliance (CL) is 200ml/cmH20 and chest wall compliance (CCW) is 244ml/cmH20.  We are then told that the value of CCW is derived from a total lung and chest wall compliance (CT) of 110ml/cmH20 and the value of CL. The missing information is that the total compliance is the sum of the reciprocals, i.e. 1/CT = 1/CL + 1/CCW, because the pressure (at a given volume) is inversely proportional to compliance.

Final thoughts

This book has a small target audience and does not require a spot on every simulation centre's bookshelf. It is probably very useful and interesting as a beginner's text for biomedical engineers and simulator designers. It will help them to understand the linkages between the final output (a clinician feeling the mannequin's pulse) and initial input (a conceptual model of the cardiac system). 

The book provides insight into the thinking and philosophy of the designers and manufacturers of mannequins which attempt to have a  realistic physiological model running in the background. Compared to the mannequin manufacturers who devolve this responsibility onto the end-user, van Meurs and colleagues have perhaps a more noble calling. However the drawbacks are that the finished product is more complicated and more expensive. It is also, as the human beings it simulates, less predictable and, if your centre owns a HPS, this book may help you understand why.