What if the robots are wrong?

There has been an explosion of the technology surrounding the health and fitness industry in recent years – particularly devices that measure and analyse the likes of heart rate and GPS data. The data and information derived from such analysis can be extremely useful – and seductive – especially to the less-informed user looking for an effective training plan, or a shortcut to a higher level of performance. This is especially the case where training is ineffective due to poor compliance, inappropriate frequency, or poor session design – the adage “any training plan is better than no training plan” is often true – this can result in high-tech solutions becoming an attractive option.

The convenience and prevalence of smartphones, smartwatches and similar hardware, along with reduced cost, has also fuelled the uptake of data recording and analysis that were previously only accessible to professional sports teams and high-performance athletes. However, there has also been a reduction in accuracy and measurement features of some of these devices, due to cost factors, mass production and the entry into the larger and less-discerning consumer market.

Many forms of technological development follow three ‘waves’ of development:

Invention: the first wave involves the creation and development of hardware that measures and/or displays certain data and parameters. For example, portable heart rate monitoring originated in the 1970s and was refined and developed over the course of many years, whilst lightweight GPS units arrived in the 1990s and took some time to become accurate, reliable and small enough to be practical for use on athletes.

Integration: the second wave sees the incorporation of measurements and scoring systems into a range of other devices (in the case of HR and GPS, into wristwatch-style units, and more recently into the array of smartphone features).

Interpretation: the third (and most dangerous!) wave is when the mass of data produced by tech/hardware is analysed by software and AI-based systems, looking for trends and suggesting what should/could be done with this personal data.

The tech involved in the health, fitness and sports industry has generally followed this process – heart rate and GPS-based hardware has formed the basis of prevalent analysis, with its origins in professional sport and the medical rehabilitation spheres. A large amount of innovation and development has been invested into developing these basic technologies, and some of their spin-offs (such as heart-rate variability analysis and systems that score speed, distance and intensity from GPS recordings), in attempts to produce viable systems that simplify the exercise journey of both the professional athlete and the man-in-the-street. However, it should be noted that these techniques are derived from data scoured from high-performance athletes, whose exercise capacity, genetics and training habits can be very different to those of recreational users.

The danger at this stage is that humans get lazy – as humans do – and start to disconnect from the measurement process, and fully trust the “automation”.

Therefore, the pitfall of over-reliance on some of these procedures (many of which have filtered into third-party applications and as separate services of their own), is that a small amount of incorrect data can lead to the erroneous analysis of training sessions and subsequent incorrect prescription of the route ahead. Simple errors can significantly impact analysis results – such as age, mass, measurements of training volume and intensity, and more complex variables such as calibration errors and over/under-reads from your connected tech. We’ve all had that bizarre 200-plus heart rate, and the 1km run at the speed of sound! Applying the exercise capacity and parameters unique to high-performance athletes can also skew the recommendations for the more average performer.

The primary remedy is to constantly review your data and recordings for obvious errors, and to be certain that analyses and calculations are reflective of your level of effort – this will keep the automation and algorithm-based recommendations as accurate as possible.

When tech fails, a wise head and a bit of rational logic will more than suffice to keep things reasonable. Use the tech and available expertise to design an appropriate training plan, but keep an eye on your stats and data, and YOU make the decisions on when to alter or adjust that plan when the numbers don’t add up. It’s a good idea to use the auto-pilot, but a better idea is to keep your eye on both your instruments and the path ahead!