Realising data and AI benefits

Data can help us understand a problem better, but only if it is collected and analysed in helpful ways. Throwing any old data at a problem is as likely to confuse as explain. The same can be said for any new data too. A structured, thought-through approach to data collection and analysis is an essential first step.

Data-driven air quality work

Air quality is an important factor in the health of passengers and staff so being data-driven is a strong benefit. Working with stakeholders at many different stations, we identified the key in air quality issues. This showed us where to focus our efforts and what methods are best suited to collect data at various sites.

We’re using a combination of diffusion tubes, reference monitors, and low-cost sensors to measure concentrations of harmful air pollutants. These include nitrogen dioxide (NO2) and particulate matter. These give us high resolution data. Monitoring over time shows us patterns in air quality issues. Since 2022, we’ve been using this data in our work with stakeholders on targeted interventions.

Evidence for climate change maturity assessments

Assessment of readiness for climate change impacts need to be evidence-based. The Climate Capacity Diagnosis and Development framework has been developed by the organisation Climate Sense over many years. We have been helping the GB rail industry to use it since 2023.

The tool allows an organisation to see its current strengths in adapting to climate change impacts, as well as any additional adaptive capabilities it may need. This helps an organisation understand what climate change is likely to mean for their own operations. It also helps them monitor their own progress. The 2023 and 2024 assessments gave foundational data for baselines and roadmaps. We are running the assessment again in 2025. We aim to increase participation to over 40 organisations. We are also looking at the data from dozens of rail organisations that participated in the previous years. This will give us a comprehensive climate change maturity assessment for the whole sector.

Applying data in tools

Using data to inform our decision making is an approach we’re using in other areas too. We’re currently working with Network Rail on developing a tool to help assess the runaway risk of rail vehicles. A pilot project to assess the tool is due to start later this year.

Useful data is already collected by industry, and the development of the Freight Wagon Condition-Based Maintenance Tool is providing new and valuable insights. This enables data-driven decisions to enhance freight wagon wheelset efficiency and safety. The introduction of the AI tool will drive safety improvements by providing early detection of wheelset deterioration. This will help prevent safety events and improve operational performance and efficiency, by optimising maintenance schedules, reducing unplanned downtime, and enhancing service reliability.

We're using this 'data combining' approach elsewhere too, for instance in the Red Aspect Approaches to Signals toolkit version 2. The long-term aim is to combine some of its data with industry performance data to give better insights.

Using AI, safely

And then there’s AI which we are using in several new tools. Our new Safe Insights tool uses AI in tandem with users to make data input easier and more accurate. For instance, when a user enters a narrative about an incident, using AI Safe Insights is able to recognise that the narrative is about, say, level crossing misuse. Safe Insights will then offer the user the relevant template. This means that the user is prompted for all the data about that event type at the time of entry, rather than waiting a month or so for the data to be reviewed. This streamlines the data entry process and hence improves data quality.

AI can also identify whether a particular piece of text is, say, a train head code, and offer to the user it at the appropriate point during event entry. The user can choose to add to the record or ignore the information. Whatever the user response, AI underpins the improved responsivity of the Safe Insights tool.

Assessing AI benefits

We’re developing an AI assessment toolkit so that the rail industry can understand the implications of potential AI tools better. Our industry doesn’t have this sort of tool yet so needs support in this area. We’ve got some advice from Frazer-Nash about what the toolkit needs to include, and how it relates to other non-rail specific guidance that already exists. But the bulk of the work is based on our expertise in rail, helping the sector explain, reflect, and understand what introducing a potential AI tool is likely to mean.

We’re proud and excited to work with all parts of the industry to find new ways to apply or combine data and AI to solve existing and future problems.