VolkerRail’s managing director, Steve Cocliff, is a member of the RSSB board, and is one of the schemes sponsors.

The new system aims to categorise close call data gathered from across the industry to identify key insights which can be used to educate the industry, allowing us all to make sites more safety conscious. It also aims to identify common incidents and trends, to reduce incident rates and highlight low priority/high frequency close calls, so that we can better understand the behavioural changes needed to reduce these trends.

RSSB is working with VolkerRail to conduct an analysis of 24,000 close call records, reported across VolkerRail from September 2019 to March 2022, using Natural Language Processing (NLP) to transform text data into a numeric format for analysis.

The NLP algorithm translates significant words and phrases found within the records into numerical representations, which are then employed as input data for a machine learning algorithm. This enables the classification of each individual record into specific incident types, such as ‘incorrect use of lifting equipment’ – with 93 high importance incidents already differentiated. Generic language, which describes multiple records across a range of incident types (e.g. “whilst carrying out”) is also being taken into consideration, and a plan is being identified to ensure this is included in the analysis, and that there is sufficient, accurate output.

Once a successful demonstration of the technical feasibility is completed, RSSB will launch the new solution within VolkerRail, and then to its wider membership group.

Reflecting on the initiative, Stuart Webster-Spriggs, VolkerRail’s HSQES director, said: “We are making great progress with RSSB to build this new system, allowing them to implement a new digital service that will not only help us analyse important data, but also see the bigger picture, by understanding trends that could pose a significant risk to the health and safety of others.

“By working with the RSSB we are also helping the rail industry implement better ways of analysing key data to improve industry safety stands, which is a great step in the right direction for a truly collaborative industry.”

Mike Briggs, director of data insights at RSSB, said: “This is an exciting and extremely challenging project due to the nature of the close call dataset, which has historically been difficult to analyse in this manner. We are very pleased to be working collaboratively with VolkerRail in developing this capability, which has the potential to be transformational for the rail industry."