Can AI make me run faster?

AI is becoming omnipresent. As a data scientist, it is at the forefront of my mind most of the time. I’m also a seriously committed runner, so I inevitably ended up asking myself if my favourite sport and past time were impacted too. I was sceptical. Surely, records were broken due to good old fashioned strenuous training; piling the miles week after week, following a schedule designed by experts based on years of experience. It’s how I started. I went through a Couch to 5k programme found in a magazine 15 years ago. After some online investigations, however, it strikes me. I’ve long ago stopped manually recording my sessions and delegated to… my watch.

It has been so embedded into my routine that I no longer notice I’m using my watch. Strapped on my wrist sits a marvel of technology, recording my biometrics every second of the day. Inside those gadgets are dozens of sensors fitted within an inch wide circle of plastic. First, of course, it’s a timekeeper with very precise chronometer, recording activities to the hundredth, sometimes thousandth, of a second. With GPS capability, gyroscope and accelerometer, it tracks how far I’ve run, the meters climbed/descended and when I accelerated or slowed down. Optical sensors monitor my heart rate and blood oxygen saturation. Insights on the intensity and quality of the training sessions, alongside derived metrics like my level of fitness and energy level throughout the day, the level of stress I am experiencing, or the quality of my sleep. All of these data points are fed to a cloud-based service, providing an impressive level of information about my health and fitness.

The difference between setting goals & gaining actionable insights

Thanks to companion apps, you can generate a training plan tailored to your ambitions. Set the distance, desired time target and a full training schedule is automatically added to your calendar. Put you trainers on and follow the workout’s instructions that have been synchronised to your watch. During the session, you’re informed of your progress, how far you’ve gone, if you’re hitting the desired speed or keeping your heart rate in the optimal range. After the exercise, you can now enjoy your runner’s high and all the benefits of a good workout. You can also review your activity in all its details.

Your tracking tool offers a plethora of performance indicators. Soon, you’re diving into a series of charts, trying to find what you did well and where you can improve. Strides length. Cadence. Heart rate zone. Running economy. Physiological load. Training monotony and strain. VO2max. Velocity of lactate threshold. You have an awful lot of facts to learn, not only about running, but about yourself, from physiology to lifestyle and a deluge of terminology, metrics, exercise types, muscle groups and medical conditions you didn’t even suspect existed. The analytics platform can compute evaluations, identify patterns and give you warnings if you miss too many sessions or struggle to hit the pace/targets. However, in my experience, rarely have I been provided with actionable insights.

How was your training, by the way? Did you find it hard or easy? Did you feel tired today? How is your mood? Your virtual coach likes to know. Inputs – that’s the key. AI is always asking for more data. It needs calibration. Add height. Add weight. Enter more data. Review the charts. Did you run with a stroller or a dog? Was it a park run or a race? Evaluate yourself. Validate the insights. If your goal was in the realm of what the algorithm has been designed to deal with, you should be able to reach your goal.

But what happens when you want to get over that? When you really want to push yourself and go faster than the tools allow?

Harnessing the power of data to achieve your goals

The problem is that most mainstream AI running solutions use historical data and proven methods, trying to match your requirements to a best-fit existing programme. If Roger Bannister had relied on AI in the 1950s, he would probably still be chasing the four-minute mile, and women would still not be allowed to run the marathon! Only humans have the ability to dream big, challenge the deemed impossible, break the conventions and change the paradigm. If you want to achieve this with AI, you need to be able to challenge it. You must become an expert in the field of running, and potentially have good notions of physiology too. You need to be able to identify which metrics matter to personalise the system and fit your reality. You’d also need some notions of machine learning to implement it. Or find someone that can.

That’s exactly what I do for my clients here at CACI, although not for running, yet. As a consultant, I embrace their goal, but first get to know who they are, where they are now on their journey, their values, strengths and challenges. I use my experience of the field and gather expertise of the industry to find the best fitting solution. As a data scientist, I can then select the appropriate technology, AI or techniques and do the data crunching, providing them with clear and actionable insights – what needs to be actioned and why – to set them on the path of success. I can keep in touch with them along the way to review and adjust their ambitions when needed as our market evolves around us.

AI can ultimately teach you how to run and improve your pace… to a point. But if you really want to get fast, you’ll also need a lot of personal experience and human expertise. Exactly as it is for data science.