Early in the pandemic, researchers started experimenting with fitness and health trackers as a way to detect COVID-19, ideally before patients noticed any symptoms, by looking at measurements like sleep levels, skin temperature, breathing rate, resting heart rate and heart rate variability (how much the heart rate changes over time in response to biophysical stress).
The US Army's Medical Technology Enterprise Consortium worked with multiple projects including Fitbit, the US Defense Threat Reduction Agency partnered with Garmin and Oura, and the NBA bought more than 1,000 Oura rings in the summer to help basketball players track their health risk.
And when Ava, whose fertility tracker bracelet is already a licenced medical product, started actively looking for researchers, the government of Lichtenstein enrolled 2,000 people in the ongoing COVI-GAPP study to see if the sleep, skin temperature, pulse rate, skin perfusion, breathing rate and heart rate variability ratio that the Ava device measures to track fertility levels can reveal a COVID-19 infection before the symptoms are obvious.
Ava tracks skin temperature directly – not all fitness trackers include a sensor for that – and it takes its measurements overnight when your basal body temperature is at its lowest, which is also the best time to measure breathing rate, resting heart rate and heart rate variability because you're resting rather than up and about. Early results from the DETECT study carried out by the Scripps Research Transactional Institute using data from wearables submitted by volunteers suggest that tracking changes in heart rate, sleep and activity levels can help spot coronavirus infections.
I'd started wearing an Ava bracelet at the end of 2019 to review how well it did at fertility tracking for someone who didn't want to get pregnant but did want to get more information about their menstrual cycle.
Having cut short a trip to Silicon Valley when conferences were cancelled due to coronavirus, and discovering that two attendees at one of the conferences I did go to had become infected, it was certainly reassuring that Ava didn't show my temperature changing from its normal monthly pattern. But I also found the device itself annoying in some ways.
Early in my testing, Ava had a glitch that reset all user accounts and the app you use to manually sync data from the device every morning throws away your login about once a month; it also doesn't store your username and password in the iOS keychain unless you visit the web page separately and log in there to force it into the keychain. At least you see every day if the app has kept you logged in because you have to open it and sync data yourself, and you have to do that after you've taken the bracelet off.
The bracelet needs charging pretty much every day, which is awkward if you're travelling frequently; charging alternate days left me with a flat device as often as not. The different states – like low power and 'memory full sync now' – are indicated by the LED turning white, pink or blue, but I found all three colours indistinguishable in daylight and blindingly bright in the dark.
At first, I had issues with the sensor taking ages to turn on when I put it on my wrist. After a month or two it started turning off in the middle of the night while I was wearing it, then turning itself on again early in the morning; the sensor buzzes when it turns on, so the early morning restart woke me up. (You can do that on purpose and use it as an alarm, but it's less helpful at 4am.) If I took it off, then it said later that it recorded insufficient data; if I left it on, it only recorded the new data and told me I'd had hardly any sleep. I tried for weeks to update the firmware; even with the bracelet and phone side by side it wouldn't find the bracelet to update even though it had just synced data. Eventually the firmware updated without me trying and the sensor and sync problems went away.
But there are still annoyances. Every morning for the first month I turned off the nightly reminder in the app to put the bracelet on; every evening I got the reminder anyway. Once it stayed off for a whole five days before coming back, which was when I gave up ever getting the notification to stay off: months later it's still on.
The app design is cluttered, focusing on the information Ava has collected rather than letting you pick the things you want to see at a glance; you can hide some of the data in the chart view but not move it onto the home screen to replace the endless fertility tips. And if you're having a busy week with a lot of early mornings, the cheery "did you know you've had less than 5 hours sleep for the last three nights" tips will make you feel extra tired – or annoyed that the bracelet turned itself off and lost most of your tracking as it did repeatedly before the firmware update. If you do click on one of the many tips that show up on the app screen, the Ava web site doesn't know that you come from the app and tries to sell you an Ava bracelet.
Even if you tell Ava you're doing fertility tracking (say you want to monitor your cycle for health reasons or to avoid surprises while you're travelling), you're pestered with tips on getting pregnant in the app. If you're using it as an aid to contraception or because of a condition like endometriosis, the obsession with pregnancy can be annoying. There's only a limited number of tips so they soon get repetitive and I couldn't find a way to turn them off. After a few cycles of gathering data, you start to get a warning that your period is coming in a couple of days; phrased not as "you might want to know", but that "your cycle is about to begin" as if you were on a self-driving bike.
Ava does learn more about you over time. After a few months it announced that I had a typical biphasic pattern – that my heart rate, breathing rate, temperature and so on were not just a predictable pattern reflecting when I ovulate but one with a common shape. At first the predictions of my period were two days early, just like the simple phone app I'd used for period tracking; but after a year the Eve app was still getting it wrong while Ava got more accurate for a while. One month it announced that my measurements showed I hadn't ovulated on the day it predicted and that it was updating its predictions – although as usual, this was phrased as being another opportunity to get pregnant.
However, the model that predicts your period based on all the data Ava collected is designed for a monthly cycle of 24-35 days and can't cope with people who have an irregular cycle, so it feels like it's only accurate when my body happens to match the rather fixed predictions, not that the model is accurately predicting my cycle. Often, it's two days early, just like a dumb tracking app; other times it's wrong by weeks.
Rather than trying to cover a wider range of fertility cycles, Ava suggests that you look at the data it collects and try to use it make your own predictions. If you're going to have to do that much work yourself, it would be nice if the app let you set rules: for example, I've noticed that my temperature peaks in the run up to my period and then drops the day before, so if the software isn't going to notice that correlation, I'd like to be able to trigger a notification when that shows up in my stats.
The wider tracking information is interesting. As a late chronotype, I was fascinated to see that the later I go to bed (so the more closely I follow my own natural circadian clock), the more REM sleep I get. I was fascinated at first by seeing how much my skin temperature varies and how that follows a cycle. But the novelty of seeing HRV and breathing patterns soon wore off and I wanted a much less intrusive app that synced automatically, had the historical data when I wanted it, and just let me know the actual day my period was starting without all the irrelevant conception advice.
In time, all that sensor data may be useful for understanding my general health or tracking the eventual onset of menopause. Given that menopause will affect roughly half the population, it's amazing how few sensors there are to help you detect and monitor menopause (precisely none that I can find). Telling me if I have a viral infection would definitely be a bonus, but with the enormous variability in human physiology, it's still not clear how broadly that can be done. The DETECT study says that sensor data plus self-documented symptoms are more useful than symptoms alone. Still, now that the early glitches are fixed, I'll keep wearing the Ava bracelet just in case and hope it branches out beyond the focus on getting pregnant.