Publications: On Google Scholar, QM Publists or Research Gate.
Current research areas are:
Medical Decision Support
System Safety and Risk
System Safety and Risk
Historically, risk assessment in safety systems has accepted models derived from knowledge; the use of data is however less common. At the same time, there is a need to expand the scope of risk models in two ways: first, to encompass the organisational and human aspects of complex (socio-technical) systems; second, to be able to condition risk estimates on specific or local conditions, to construct a risk profile (by location or installation) rather than merely an overall aggregate.
Risk Models
Working with George Bearfield and others at RSSB, I have proposed ways to extend fault and event trees to create risk profile. I have applied this to models of both high severity, low frequency incidents and to models of lower severity, high frequency incidents, where the model can be partly learnt from data.
Selected Papers
Marsh, DWR; Bearfield, G; Generalizing event trees using Bayesian networks Proceedings of the Institution of Mechanical Engineers, Part O: Journal of Risk and Reliability, vol 222, no 2, p105-114, 2008
Marsh, William; Bearfield, George; Representing parameterised fault trees using Bayesian networks Computer Safety, Reliability, and Security, p120-133, 2007
Hazards in Complex Systems
Hazards are a central idea in practical risk analysis and in complex systems it is naturally desirable to structure the analysis by subsystems and to reuse elements of a hazard analysis from related systems. Working with Rune Winther at COWI, I have reviewed the confusion that surrounds hazards in complex systems and proposed a structure for coherent hazards in complex systems.
Selected Papers
Presentations
- Why Risk Models Should be Parameterised. MASR workshop, 2009
- Decision Making from Data: Causes and Uncertainty. INCOSE UK Chapter, Rail Interest Group, 2011
- Change and Safety: Decision-Making from Data, RRUKA Conference, 2012
- Hazards, accidents and events - A land of confusing terms, ESREL 2013
- Improved Decision Making for Maintenance using Data, RRUKA 2013