WADA eyes AI, big data analytics to fight illegal doping

01-Jun-2020 - Last updated on 01-Jun-2020 at 15:52 GMT

The pandemic has shut down in-person anti-doping testing in many countries, placing a renewed focus on artificial intelligence.

Just over two years ago, World Anti-Doping Agency director Olivier Niggli revealed that discussions on the use of artificial intelligence were underway.

Now, with a pandemic at the forefront, Niggli confirmed that pilot projects are being planned for later this year to fight illegal doping in sport.

At what cost? WADA is funding four AI projects in Canada and Germany that would help indicate signs of drug use. The three projects in Canada cost WADA about $425,000 over two years, with matching funding from the province of Quebec's research funds. The project in Germany costs $60,000, according to WADA.

Red Flags

The AI tool would be used to flag suspicious athletes and make sure those who raise red flags get tested.

“When you are working for an anti-doping organization and you want to target some athletes, you look at their competition calendar and you look at their whereabouts, you look at the previous results and so forth," said WADA senior executive director Olivier Rabin. “But there is (only) so much a brain can process in terms of information."

A complex system

With a lack of financial and human resources, WADA can only comb through so much data. The organization is hoping that by employing sophisticated algorithms to spot anomalies, they can zero in on the more questionable athletes.

Dope tests are currently used to analyze an athlete's blood or urine sample to detect performance-enhancing substances. They also investigate several biomarkers such as red blood cell count or testosterone levels in the “biological passport” program to detect the effects of using substances such as the blood-booster EPO.

Currently, athletes' results in competition are not taken into account, but Rabin indicated it is a possibility in the future.

Taking it up a notch

The organization wants to make EPO and steroid detection more precise by using AI to track patterns between the biomarkers and cross-referencing them with other data.

Machine learning, an application of AI, can be used to teach the system 'dirty' and 'clean' profiles, data likely missed by human observers.

Privacy concerns

Rabin revealed a 'global' project in Montreal which could predict the risk of doping by evaluating data from a wider range of sources. This might include information on athletes’ locations, which they are required to file. However, their personal data and the cities where they live and train will be anonymized due to privacy concerns.

“It's been fairly complex discussions ... to try to find a balance between, you know, again, protecting data, protecting individuals and making sure that you can still reveal the potential of AI, if there is any,” Rabin said.

Unprecedented times

In response to coronavirus, most international anti-doping agencies stopped testing in mid-March.

NutraIngredients-USA recently reported on Project Believe 2020, a virtual drug-testing program started by the United States Anti-Doping Agency. In early April, USADA invited 15 elite athletes to volunteer in the pilot program to conduct urine and blood tests in their homes on their own while being observed remotely by phone and video conferencing.