AI to Fully Unleash Humanity’s Secret Superpower -Running
If animals could participate in the Olympic games, homo sapiens would end up at the lower end of the medal ranking. Imagine Usain Bolt VS. Cheetah or Michael Phelps vs Sailfish. The only event we might win a medal in is the marathon because the human body is tailored for long-distance running. In 42.195km we can outrun nearly any species:
1. Ostrich 45 min
2. Antelope 45 min
3. Camel 60min
4. Sled Dog 70min
5. Human 120min
The primary reason why humans can outperform most of the species is the efficiency of our cooling systems — our ability to sweat. On the downside our major disadvantage is that we don’t have an in-born endurance; we have to train. The performance difference between a slow and fast Camel is not significant when compared to humans. The fastest human marathon runner finishes a bit above 2h and it takes an average runner 4h. The majority of humans are not even able to run the distance at all.
The complexity of an effective training plan
To optimize performance we need to train, but sports performance is a tricky, deceiving monster. It is complex to understand the variables and even harder to control them in a practical way. The complexity of the human body is the major reason that sports sciences understanding about the effects of exercise and training on health and sporting performance is at the very beginning. Another challenge is that every human is different, not only from a genetic perspective but also age, gender, race, and psychological factors that need to be taken into consideration. Much of sports performance is dictated by the right timing, building up endurance and strength, by avoiding over- and under-training, which both lead to injuries. Taking all those factors into consideration, it is common sense that one size fits all training plans are just an accepted evil, due to the absence of better solutions. In other words, we aren’t talking about absolute sports performance, but personal performance relative to genetics and many other influencing factors, a complexity, which can be handled with a static training plan, and goes far beyond what a human brain can understand. Researches have shown that some individuals improved by less than 5%, others improve by 30%, following the same training plan.
Data-Driven Insights into your training
What if, a machine can help in analyzing millions of marathon results and relate it to training data of individual athletes, and ultimately provide data-driven recommendations to the athlete and coach? Acknowledging that today’s running trackers are limited to a handful of metrics, this wouldn’t replace a coach, but it can help to answer some fundamental questions in a personalized way:
- How many hours per week to train?
- What’s the most effective pace?
- What’s the longest training distance to run?
- How many days before the race the longest run should happen?
- How many days of taper?
- What’s my predicted finish time, based on my training?
Also, those recommendations look fairly basic. Today’s reality is that sports science has no solid answers for individual athletes, and that’s why training-plans shows significant differences and contradicting information such as the relevance of training volume — the recommendations vary from 3–10 hours per week, and it remains more of a gamble when you start to taper.
Can an Algorithm do better than a human coach?
Acknowledging that such an algorithm doesn’t exist and today’s digital training data is limited to distance, time and heart rate, it won’t add much value to the elite runners or replace a human coach for the advanced runners. However, it has the potential to change the game for the big mass of first-timers and beginners who would otherwise download a $50 plan from the web. A machine is capable to learn quickly which training practices have a significant impact on the race result or injury risk and could even predict your finishing time. It means an athlete could get the exact number of miles and pace to run in the next week in order to optimize the race result and decrease the injury risk. Another major advantage of data is that a machine considers days-to-race, which is a major factor. Today’s training plans are only touching days to race on the surface and can’t adjust in case the runner could not follow the plan.
