I’ve been using an Apple Watch and Apple Health almost since they were introduced. Before that, I used various other tracking tools. But a combination of privacy concerns and being able to consolidate my activity tracking to a single service has resulted in me putting my health eggs in Apple’s basket. But when using a service for things like fitness tracking, it’s important to understand the difference between things that are actually being measured and those that are being calculated.
Distance and algorithms
The Apple Watch (and pretty much every other activity tracking watch out there) ships with a GPS receiver that allows it, within a quite small distance, to record your position on the planet. You’d think they would all record exactly the same distance when taken on the same journey. So why is it that two people running, cycling or walking side by side can record different distances? The answer is in the software.
Your watch doesn’t monitor your movement constantly. If it did, the GPS would wipe your battery out quite quickly. Instead, your position is recorded periodically and software makes predictions about how you go between each point. The exact distance recorded will depend on how often the watch records your position.
Look at the image below. The distance between the start and end on the left will be shorter than the distance between the points for the route on the left. If this was a journey being measured by a watch, the route on the left represents a route where the GPS records the position less frequently than on the right.
What the software might do, if it can use other data such as a track or road, is predict where you’re moved between each read of your coordinates. But there will still be small differences depending on whether to algorithm that fills in the blanks uses the middle of the track or road or takes the shortest distances through bends.
If you’ve ever run in a fun fun or road race, you’d know that your tracked distance may vary from the offical distance depending on whether you ran the tangents on each turn or took the corners on the wide side.
Measures of fitness
Aerobic fitness can be measured in many ways but the one metric that stands out most is VO2 Max. VO2 Max is a measure of how well your body can use oxygen. As you train for progressively longer and more intense periods, you can increase your VO2 Max through better lung capacity, improved heart efficiency and development of your circulatory system. As your body gets better at using oxygen, your VO2 Max increases.
Accurately measuring VO2 Max requires someone to be placed on a treadmill, exercise bike or some similar device. The subject is then connected to a bunch of devices that accurately measure the amount of oxygen being sucked in when inhaling and being expelled when exhaling, while monitoring heart rate and other metrics.
Clearly, this is not what is happening when Apple Health (or any other app or website) predicts your VO2 Max.
Instead, an algorithm takes into account your age, weight, gender, heart rate and other data to predict your VO2 Max. The prediction is based on data that has been accurately collected over many years to build models. However, like all machine learning models, the accuracy of the algorithm’s prediction is dependent on the quality of the data used to train the model. And there are also different prediction models.
Importantly, a lot of the data we have regarding VO2 Max comes from collecting data from certain cohorts of people. For example, a lot of the data comes from measuring athletes, military service people, and people of specific ethnicities where genetic factors are also a factor.
Apple Health uses a sub-maximal prediction model rather than Peak VO2. This means Apple’s model can use all your exercise data and not just the data from when it deems (and that’s another prediction based on age, weight, gender, etc) your heart rate is at its peak. And even if two different platforms use the same type of model, they can offer different results for a predicted VO2 Max as they are, most likely, suing different algorithms to make the prediction.
My advice is to not look at measures such as VO2 Max as an absolute number but rather, track the trend to see if you’re improving, holding steady or need to step things up a bit in your training routine.
Sleep quality
Sleep tracking is one of those measures that I find really interesting. Sleep is, despite a mountain of research, still a huge mystery to scientists. But there are some things we definitely do know such as sleeping and waking at about the same time each day seems to be conducive to feeling better and more energetic and alert, and that we all have our own sleep rhythm. We all go through cycles of light, deep and REM (rapid eye movement) sleep when we are asleep.
Sleep tracking uses algorithms to make predictions about what sleep stage we were in and how long we spent in that state. This can be based on movement, sound, heart rate, respiration rate (which is predicted using an algorithm) and models using data collected during specific sleep studies.
The data being collected by an Apple Watch can give us lots of useful insights into our health, fitness and general well-being. But it’s important to understand that the numbers are not all absolute truth. They are calculations based on measurements and algorithms that are built from models that use data collected through other means. I have friends whose predicted VO2 Max from an Apple Watch is lower than men but who clearly have better endurance than I do. Why is that? It’s because the algorithm is making its best guess at VO2 Max and, sometimes, the margin of error in that guess results in some curious results.
This doesn’t mean the metrics are bad. But rather than treating them as absolute truth use them to track progress.
Anthony is the founder of Australian Apple News. He is a long-time Apple user and former editor of Australian Macworld. He has contributed to many technology magazines and newspapers as well as appearing regularly on radio and occasionally on TV.