The Russian doping scandal, and the recent case of American world champion sprinter Christian Coleman – who committed three whereabouts violations in 12 months, but was cleared on a technicality – have highlighted concerns with the current anti-doping system. Prominent anti-doping scientist Don Catlin has said that fewer than 10 per cent of dopers are caught.
You’re going to look at the blood profile of the athletes over the past few years and whether the athlete has made a habit of changing his or her whereabouts shortly before the mandatory hour for testing.
Dr Olivier Rabin
“Artificial intelligence offers you one big benefit: using a big batch of data that the human brain cannot process because it’s too much for human capacity,” said Dr Olivier Rabin, WADA’s director of science. “We want to use artificial intelligence for improving our targeting capacity. Identifying suspicious correlations could be an indication of doped athletes.”
Last month, WADA announced that it was funding three projects to explore the use of AI in doping, all undertaken by independent researchers. “Athletes are smart in how they dope, you need to be smart in how you apply anti-doping tests,” Rabin said.
“AI will help us in the recommendation to test this particular athlete at this particular time to attempt to reveal doping.”
Should the research projects prove as successful as hoped, Rabin says AI could be used within 18-24 months. “It’s going to help us focus on the suspicious data and the suspicious athletes’ profiles,” he added.
Targeted testing is far more effective than random testing, but WADA still believes that not enough tests are sufficiently focused. Using artificial intelligence could help spot new patterns typical of dopers – assisting its investigations.
“We want to see the correlations that are unknown to us,” Rabin said. “There is a correlation that we could find between some of the aspects of the suspicious profile and between the population of some of the athletes that we could identify as being doped. Revealing those correlations with AI is something we would like to automatically apply in our system, once validated.”
Patterns typical of dopers could be revealed by AI being able to analyse large sets of data quicker than human beings, Rabin said.
“You’re going to look at the blood profile of the athletes over the past few years and whether the athlete has made a habit of changing his or her whereabouts shortly before the mandatory hour for testing. There are different sets of data you can combine and correlate. If you have a strong correlation revealed by AI, then all the athletes who have those patterns in their profile, you should aim at testing them two hours before the mandatory window.”
Rather than using AI as sufficient proof to catch dopers, it would more likely be used to focus drug testing on suspicious athletes at the times they were most likely to dope. This would build on work done by the athlete biological passport, introduced a decade ago.
Meanwhile, genetic drug testing could be introduced to sport in time for next year’s Olympics, the president of the International Olympic Committee said.
Two months after the Telegraph revealed a breakthrough was close, Thomas Bach said it may be ready for Tokyo 2020.
He also confirmed it could be accompanied by dried blood spot testing (DBS), pending WADA approval. “This new approach could be a ground-breaking method to detect blood doping, weeks or even months after it took place,” Bach said in a speech at the start of the World Conference on Doping in Sport in Katowice. “If approved by WADA, such new gene testing could be used already at Tokyo 2020.”