Dr Martin O’Reilly of UCD and the Insight Centre for Data Analytics. Image: UCD
Dr Martin O’Reilly of UCD is working to bring the latest sports analytics technology to a point where even a Sunday league team can access it.
Dr Martin O’Reilly of the University College Dublin (UCD) School of Public Health, Physiotherapy and Sports Science, and the Insight Centre for Data Analytics started working in his current field after completing an undergraduate placement at Shimmer Research in 2013.
The following year, he undertook a PhD in how wearable sensors could augment strength and conditioning training. His subsequent PhD work has been published in 18 peer-reviewed papers and he will be a principal investigator for a planned start-up called Output Sports, which aims to develop sports technology that is cost-effective, scientifically valid and user-centric.
What inspired you to become a researcher?
I don’t think there is an exact moment where I decided I wanted to be a researcher, but I’ve had a long-term desire to understand as much as possible about the complexity of sport and exercise.
I have always been sports-obsessed and passionate about technology and problem-solving. I was recently clearing out the attic in my family home and I found a copybook from when I was about seven or eight years old, where my hobbies were electronics and sport. Apart from the fact I don’t do three types of swimming any more, not much has changed!
Can you tell us about the research you’re currently working on?
After finishing my PhD last year, where I developed and evaluated wearable sensor systems for gym-based exercise analysis and feedback, I am now working collaboratively with sports scientists and computer scientists to develop wearable sensor systems that can analyse a diverse range of athlete performance metrics and injury risk factors.
In particular, I am working with Dr Darragh Whelan, a sports medicine specialist and physiotherapist, and Julian Eberle, a talented programmer, to develop athlete analysis systems for coaches to offer a new level of practicality, portability and data integration compared to existing methods.
Our key interest is maximising the amount of valuable data that can be extracted for sports coaches from a single wearable motion sensor. We approach this challenge by collecting unique datasets, which synchronise wearable sensor data with state-of-the-art biomechanical research equipment, such as optoelectronic motion capture and force plates.
My work involves applying signal processing and machine learning to the wearable sensor data so that it can produce meaningful insights for sports practitioners akin to those that can be received from the biomechanical research equipment.
In your opinion, why is your research important?
Athlete testing and tracking currently involves the use of cumbersome, heavy, bespoke equipment, which measures a small subset of fitness attributes. At an elite level, this makes testing and tracking resource-intensive from a cost and time standpoint. At a sub-elite sports level, many of the existing technologies are prohibitively expensive and impractical for coaches. This means they become reliant on subjective techniques for athlete analysis such as visual analysis and self-reporting.
The research I do is important because it mitigates the above issues, allowing coaches to gain a true understanding of their athletes in a more practical, efficient and automated fashion than previously possible. This empowers them to optimise athlete performance.
What commercial applications do you foresee for your research?
Currently, I am working as principal investigator on an Enterprise Ireland commercialisation fund, which will hopefully lead to the establishment of a start-up company, Output Sports, next autumn. At the moment, our team is taking part in the UCD VentureLaunch Accelerator Programme at NovaUCD, which is helping us refine our commercial proposition and strategy.
Additionally, for the first time, we are completing user tests with a selection of elite sports teams from Ireland, the UK, Canada, the Netherlands and the US. It is very exciting to see how the research we have been completing for years may add real value for sports coaches at both the elite and sub-elite levels.
What are some of the biggest challenges you face as a researcher in your field?
Developing new sports and exercise technologies requires a very broad range of skills including sports science, engineering, data science, psychology, anthropology and so on. A major challenge is building the right support network of experts to complete research that advances the field.
One of the most enjoyable things about the research I do is the amazing diversity of experts I get to work with and learn from day to day, such as Darragh (sports physio) and Julian (developer). Over the years, it has been super to get support from academics ranging from signal processing and machine-learning experts such as Prof Tomas Ward (Dublin City University) to sports science experts such as Dr Eamonn Delahunt (UCD).
I am also lucky to work with Dr Catherine Mooney (UCD), who was a physiotherapist before becoming an associate professor of machine learning, and Prof Brian Caulfield (my PhD supervisor) of the Insight Centre for Data Analytics and the former head of the School of Physiotherapy at UCD.
In short, I believe passionate, interdisciplinary teams are essential in overcoming challenges producing impactful research in the field of sports technology.
Are there any common misconceptions about this area of research?
Lots! One of the major things is that people perceive the most important part of developing a wearable sensor system for exercise analysis is creating the actual hardware device. In truth, these consist of relatively simple technologies such as accelerometers and gyroscopes, which are simply higher-resolution and more configurable versions of the sensors that exist in almost all smartphones.
The true challenge is in extracting meaning from the signals these sensors produce. To do this, rich datasets must be collected, and advanced signal processing and machine-learning methods must be utilised. The data analysis pathways that are applied to the wearable sensor signals is where most of the research in wearable sensor systems takes place.
Are you a researcher with an interesting project to share? Let us know by emailing editorial@siliconrepublic.com with the subject line ‘Science Uncovered’.
