The Accelerometer: So Much More than a Step-Counter

How does the industry currently use accelerometers?

First, we need to answer the following: what is an accelerometer? An accelerometer, most fundamentally, is a device that detects changes in an object’s speed over time. In our smartphone, the accelerometer resides in a tiny chip and detects movements of the phone; these movements could be suuuper slow or incredibly fast, and could be tiny (even sensing vibrations difficult to perceive by our eyes) or large (throwing your phone across the room). 

Importantly, our modern smartphones are sampling at high frequencies (>100 Hz). Because voluntary human movement typically occurs at low frequencies (< 10 Hz), our smartphones are perfectly suited to measure our bodily rhythms. 

Traditionally, accelerometers in smartphones (and wearables) are used to transform movements into some sort of insight. I’ll give you a few examples:

• Step counting: unique, repetitive patterns of walks and runs can be identified in acceleration data.

• Calories burned: intensity and duration of movement is often combined with your personal data to calculate energy expenditure. More vigorous movement (big changes in acceleration = higher calorie estimate).

• Active minutes: time spent in movements that exceed a certain intensity threshold (like brisk walking or running) can be determined by the accelerometer's data.

• Sleep tracking: Prolonged minimal movement indicates sleep. Sleep stages can even be predicted, as amounts of movement change across stages.

Going beyond the good ol’ conventional metrics

So, maybe I’ve convinced you that accelerometers are pretty powerful. But, what if I told you that we’re just scratching the surface? These phone-based accelerometers are so sensitive that they appear to be able to detect heart beats, just by holding the phone (check out this study and this review). This opens the door to tracking heart rate and heart rate variability with an incredibly accessible device that you already own (of course, standardized procedures and data processing will be necessary), but wow, talk about an amazing use-case.

Indeed, the accelerometer is pretty great, but one frustrating part of the wearable/wellness/fitness space is this obsession with cardiovascular and physical activity metrics (as I’ve talked about before). At umo, we’re using the accelerometer for a completely different sort of metric.

We are using the accelerometer to focus on tiny, otherwise invisible micro-movements of your hand: low-amplitude tremors (maybe you’re familiar with this word from Parkinson’s Disease, but everyone has tremors, so don’t worry). Changes in these micromovements are hypothesized to be related to nervous system function, specifically fatigue (check out data from speed skaters, youth athletes, non-trained men, and healthcare professionals). By analyzing frequency based data, we can infer changes in your nervous system. To expand on this just a little bit… most people look at time-domain data: in the case of acceleration, this data shows how your body moves and at what intensity. In our approach, we see the frequencies inside that motion - the rhythms that reflect how the brain and muscles communicate. 

Is this evidence-based?

Because we know that our nervous system communicates via frequency channels (sort of like a radio), that muscle fatigue alters brain activity (evidence here), and that our brains and muscles communicate directly (via the corticospinal tract; read about such communication during fatigue here), we believe it's possible to pick up on neural changes using hand tremors. My own PhD work shows that there is incredibly strong coupling between frequencies rhythms in the brain and hand muscle (check it out).

Our goal? To use (proxy) neural data for three things: precise training, efficient recovery, and proactive injury prevention. It’s time to make the nervous system accessible to everyone.