‘This has got away from us’: Experts call for oversight on ‘wild west’ of athlete data collection

When Jason Weber began working as a strength and conditioning coach in the 1990s, the collection of athlete data was done with a stopwatch and a pen.

In the decades since, the sporting world has changed dramatically. Turn up to a training session of any professional team – and many junior ones too – and you will invariably see staff attaching GPS units behind the necks of players and a gaggle of analysts on the sidelines monitoring laptops.

And that’s just the surface.

“I think we probably got the first GPS units around 2004-2005, and that’s when it really started to escalate,” Weber said.

“And from then on, it is probably what they call the ‘quickening’. Things have just escalated to the point where some athletes are now wearing devices which take biometric data 24 hours a day.”

Today’s athletes are so routinely measured, timed, filmed and tracked on sensitive information such as sleep and menstrual cycles that data production is now as much a part of daily life as a drop-punt or a defensive drill.

But whether all that data is actually useful, and what happens to the oceans of information after it disappears into a computer system, is at the core of a new discussion paper prepared by the Australian Academy of Science, in collaboration the UWA Minderoo Tech & Policy Lab.

A Melbourne Demons player is fitted with a GPS unit.Credit: Simon Schluter

Titled Getting Ahead of the Game: Athlete Data in Professional Sport, the paper was created by a 12-member panel including experts in sports science, medicine, law and governance, artificial intelligence, former athletes and experts in player rights representation, and it calls for an overdue look at the “wild west″⁣ of data collection in sport.

The paper found the collection of athlete data is “expansive, invasive and unchecked”, and highlights an “urgent need” to improve standards for the collection, storage and use of personal information in professional sport in Australia.

It also finds little evidence that much of the data currently being collected is actually benefiting athletes and is instead being harvested and – without the athlete’s understanding – shared with third parties in the hope tech companies and machine learning can glean some insights.

Wallabies players Nathan Sharpe,Ben McCalman and James O’Connor fix GPS tracking units in Durban in 2011.Credit: Getty

“There has been this massive expansion in the scale and dimensionality of athlete information and it is unregulated, with no safeguards and minimal benefits to athletes,” UWA Professor of Law and Technology Julia Powles, one of the paper’s lead authors, said.

“It is just an acknowledgement this has got away from us, and it’s just like all of our interactions with digital platforms, we are so dependent on them and feed the beast and never worry about what the implications of that are, and who is watching the watchers.

“The presumable intention of data collection is improving performance, preventing injury, great health outcomes. But what we came to realise is that a lot of the data doesn’t have a good reason to be collected, and to exist. I have worked in other sectors where that becomes a huge problem.”

Commonly collected data from athletes comes from GPS tracking units, heart rate monitors, accelerometer devices, video monitoring and self-reporting logging apps that keep track of an athlete’s diet, sleep and wellness.

Heart rate devices lined up for athletes at a Geelong training session.Credit: Getty

It has yielded extraordinary amounts of data but, as Weber describes, no-one put much thought into where it would all go and who would access it. The AAS paper argues the current data practices often exceed limits of what is “reasonably necessary” and athletes often have limited knowledge of where it goes.

“There are documented cases of sports having huge bodies of data that they don’t really know what to do with, and then they farm them off to larger organisations with higher computational skills, like your Google and things like that,” Weber said.

“I was part of the wild west, I was part of ‘we will just get on with doing it’. And no one understood what you’re doing so no one asks questions. We have to better than that. I interviewed one team sport athlete with lots of experience and he said: ‘Oh that group I worked with, they have about five years of my sleep data, I wonder what happened to that?’”

Jason Weber in his time withe Fremantle Dockers,

Powles said sports have to recalibrate the privacy of athlete data, which most treat under a “performance″⁣ category and not as personal health information.

“That is an illusion. Legally, you have non-personal information or personal information, and then personal information that is in any way used to reflect an individual’s health or make projections about their health, which is why all this stuff gets collected to be able to do load management, measure fatigue, recovery and that sort of thing. As soon as anyone is using it for that purpose, it becomes health information,” she said.

“There are a proliferation of systems that no one really understands, they have these athlete management systems, third-party tech, just pushing stuff up into the cloud, it’s going offshore, it’s going all over the place. What I think a recognition of this information being health information forces is much more assessment in ‘why are we actually collecting it?’, and be more considered and justified about why the information is being collected. And then to be able to manage it properly.”

Dyson Heppell of the Bombers inserts a GPS device on Dylan Shiel.Credit: Getty

Powles said even sensitive information provided by female athletes such as menstrual information was found to be accessible by a range of different people in organisations.

“To collect it and share it to all and sundry, just because you can, is not at all in the interests of the athletes, or their performance outcomes,” she said.

Weber, a former Wallabies and Fremantle Dockers high-performance coach who contributed to the discussion paper, believes colleagues in his industry should all be “more accountable” and first make a case to collect data from athletes for a “viable” purpose.

“There is a section of all the promise of machine learning, where there is this idea lets collect it all for a rainy day and maybe some future tech,” Powles said.

“I know from talking to those tech guys that players and organisations are coming to those companies saying can you find something that all of sports science has never been able to find? It’s looking for needles in haystacks about what might influence performance.”

Powles is hoping the discussion paper will serve as a “sense check” after two decades of wild growth in the sports data field and prompt sporting bodies, players representative groups and the government to address necessary change in governance and regulatory systems, with possible additions of independent governing bodies, structures to monitor of future developments in technology and even an ombudsman.

The paper also recommends a sport-specific privacy code, going beyond the limited scope currently provided by the Privacy Act for the sharing of athlete data to third-parties.

“We need an athlete-focused approach to data collection and understand they’re people,” Weber said.

“We need to inform everybody that we are measuring things on people. We are taking information about a person. It is not just a commodity.”

News, results and expert analysis from the weekend of sport sent every Monday. Sign up for our Sport newsletter.