Wednesday, January 16, 2013

An Upcoming Workshop On Benefits Management In Health IT - Late February 2013.

Workshop Objective.
To understand how to manage the benefits of health IT and the fact that this depends on understanding the dynamics of learning to improve health systems.
There are places available for the 2 day and 5 day workshop in Sydney.
The title is: Dynamic Modelling: What, Why and How? Examples from New Technology Adoption, Infectious Disease and Health Care.
Two world expert practitioners and trainers Nate Osgood, ex MIT, and Andrei Borshchev, from St Petersburg, will be visiting from 18th to 26th  Feb. They specialise in combining different dynamic modelling methods including Discrete event, System Dynamics and Agent Based Simulation in Health and other Industries. There is a 2 day overview or a 5 day practical training option, and a likely choice modelling workshop the following week (Mon 25th and Tues 26th). The current details (including a downloadable flyer and online registration link) are on a temporary wiki page at
Please email  for further details
Seems to me that this may be very interesting for many in both the private and public sector - especially those involved in policy formulation and leadership.
I do hope you and your colleagues can attend this exciting event.


Bernard Robertson-Dunn said...

At the risk of offending anyone, I think it is useful to draw some boundaries around this course.

The workshop is training in a particular modelling product, AnyLogic.


The tool was named AnyLogic, because it supported all three well-known modeling approaches:

-> System dynamics,
-> Discrete event simulation,
-> Agent-based modeling.

From what I have read of AnyLogic, it is a useful tool in certain areas. e.g. modelling how an A&E department might work, or to examine queuing issues in a busy practice.

However, the modelling approaches that AnyLogic supports are not the only mechanisms for Dynamic Modelling, however my greater criticism is that these techniques of dynamic modelling are best used in quantitive modelling, i.e. where concepts are very well defined and can be accurately measured, numerically.

Benefits, management and the whole area of health (except, maybe in health statistics) are full of vague and ill (excuse the pun) defined concepts. The danger is that the use of modelling techniques suited for science and engineering (which dynamic modelling is) is not suitable for soft sciences such as management and politics.

This issue becomes very apparent if the people doing the modelling were to start with defining a taxonomy and measuring regime agreed by all parties.

I very much doubt that AnyLogic would be of much help in policy formulation, value chain analysis or benefits modelling.

This is where this type of modelling overlaps with the PCEHR - information. I still believe the biggest weakness of the PCEHR and NEHRS is a lack of understanding of information and its behaviour - and no-one has disabused me of that opinion.

Once again, this is only my opinion and I'm happy to be corrected.

Geoff McDonnell said...

“All models are wrong, some models are useful”. As George Box commented, Bernard. The workshop does focus on simulating quantitative models and linking dynamic models with data, including from mobile devices.

The modelling process has a large qualitative stage which covers many of the points you mention, particularly the need for conceptual clarity and testable hypotheses. In fact, James Hansen the Climate Change modeller wrote, “A model is a word for people who can’t spell hypothesis.”

The fact is, these methods have been successfully used in the areas you mention, by the course instructors and in particular system dynamics models are common in management, business and policy by John Sterman, George Richardson, John Morecroft, David Lane, Jack Homer, Bobby Milstein and Gary Hirsch, to name a few.

I do agree that these problems are too important to be left to one method. Hence the use of a multi-method tool and a focus on hybrid models. Having been an information modeller half my working life and a dynamic modeller the other half, I’m sure they can usefully complement each other. As my friend Gene Bellinger puts it, It’s the VALUE of AND.

Bernard Robertson-Dunn said...


Re: “All models are wrong, some models are useful” and “A model is a word for people who can’t spell hypothesis.”

Those statements reflect a very narrow view of modelling.

Models can be used to document things and define behaviour - most modern cars behave the way they do because of the models embedded in them. Similarly with process control models. These models are neither wrong nor hypotheses, they define reality.

In information systems, data models are used to define the behaviour of information. They are also used to define relationships and data flows. Some models may not reflect reality, but that is not because of a failure of the model, more a failure of the modeller.

My point, in the context of the PCEHR, is that not enough attention has been paid to modelling and understanding the information. Modelling that needs to be definitive and accurate, certainly not wrong or hypothetical.

A similar discussion, on the difference between quantitative and qualitative modelling, has been going on in Gene Bellinger's System Thinking World on LinkedIn:

Maybe further discussion on modelling should continue in that forum.