Exclusive:Successful Los Angeles crime prediction trial to be copied on Scottish rail network
Remotely-operated drones could be deployed to spot incidents on tracks
A trial to predict crime on Scotland’s rail network is to be launched following a project which scored a high success rate on the streets of Los Angeles, The Scotsman has learned.
Network Rail is teaming up with academics for the pilot scheme which could be deployed during the Commonwealth Games in Glasgow in July.
It hoped the forecasting could cut the increased disruption to trains from trespassers during major sporting events, such as when drunken fans miss their trains and walk along tracks.
British Transport Police figures showed there were 635 trespass incidents on Scotland’s railways between April and September last year, up 17 per cent on the same period in 2024, which caused 228 hours of train delays - or 9.5 days’ worth.
Remotely-operated drones, known as “drones in a box”, could be deployed to spot incidents on lines.
Sean Leahy, an air operations project manager at Network Rail, said an associate professor at Kingston University in London had been “working really successfully” with the City of Los Angeles on crime prediction.
Dr Naru Shiode, whose work includes researching crime patterns, got involved after being caught in train delays around major rugby matches at Twickenham.
Mr Leahy told the Unlocking Innovation rail conference in Glasgow: “When I saw the [LA] presentation, I was gobsmacked because I did not think you could predict crime - it’s completely random.
“They took all the data from all the crimes they had and their locations of where they happened, the time of day and what else was happening that day such as weather and socio-economic things, like whether lots of people had been laid off that day, or there were big sporting events on.
“They started to predict crime and were getting between 65 and 85 per cent accuracy of predicting where crime was going to happen, down to the point of when it was going to happen.
“We can’t really predict trespass and vandalism on the railway. If we could use what they have successfully done in Los Angeles, even if we got it 50 per cent right, it’s still 100 per cent better than what we are doing now.
“It would allow us to make better decision making.”
A spokesperson for Network Rail said: "Safety is our top priority and we're committed to reducing any incidents of crime on Scotland's railway.
“We are always keen to explore new technology that can help us do that, looking to success stories elsewhere.
"We work closely with partners, including the British Transport Police, to ensure any measures we consider adopting are effective, proportionate and in the best interests of passengers and staff.”
The track and signalling body said it was not currently using any AI crime prediction tools and plans had yet to be confirmed to progress such work in Scotland.
Phil Campbell, ScotRail’s customer operations director, said: “We are absolutely committed to doing everything we can to ensure the railway is safe for our customers and our staff.
“We will continue to work closely with the British Transport Police and industry partners to reduce incidents of trespass and vandalism, and make sure people feel safe while they’re travelling on Scotland’s railway.”