Hugh’s Sleep Diary: Week 2 - I'm drowning in all of the data

It's time for bed, and I feel ridiculous. I have two devices strapped to each arm, all of them poised to start tracking my sleep, although you'd think I was about to embark on important work for NASA.

My real mission was less thrilling, but deceptively complex: to find out how well I'm sleeping, and improve it. Task one was to track my sleep for several days using different devices, and see what, if anything, I could learn from the information. It's a very basic first step, but an important one; many fitness trackers and smartwatches provide sleep tracking, but inferring useful information and applying it to your lifestyle is a different matter. I also didn't want to start dramatically shifting my sleep patterns for the test yet, better to study my patterns as they are to help me understand what I should improve.

I wanted to use a variety of devices, so I went for a Pebble Time, two different Fitbits (Blaze and Surge), the Fossil Marshal running the Sleep as Android app, and the Apple Watch running the Sleep++ app. I have others to try, but I wanted a bit of consistency for a few days, and this gave me a mix of more dedicated and less dedicated sleep trackers. Also worth noting: I haven't yet started using sleep-cycle based alarms, so I've just been setting my alarm for whenever I want or need to wake up.

Read this: Sleep trackers tried and tested

When it came to accuracy, there was definitely some variation. I don't want to linger too much on specific differences, but it's worth pointing out that accuracy does tend to fluctuate, as you may have noticed from tracking with your own devices.

Last week, my sleep expert, Jamie M. Zeitzer, told me I should try tracking some bad nights' sleep and see how well the data is picked up. I've been paying more attention to my patterns the last couple of weeks, and it's become clear that when I sleep in later on weekends I find it hard to fall asleep the next night. That might sound obvious, but on Sunday I also did an extensive workout and really should have been ready to pass out in a flash. Yet I was still feeling very much awake when it came to bedtime, and it took an hour and a half for me to drift off.

It was a good chance to see which devices best interpreted what was happening. The Fitbit and Pebble correctly didn't start tracking my sleep until I'd actually nodded off, while the other two, which relied on me telling them when I had got into bed, were doing more guesswork. The Sleep++ data picked up that I was restless but any still moments counted towards sleep. When data was analysed by Apple Health it corrected its error, concluding that I fell asleep at at 1:57 and adjusting the total to 7h 19min of sleep. That post-sleep analysis immediately showed how important it is to not only rely on movement monitors alone.

Meanwhile, Fitbit said I was asleep at 1:49am, and the Pebble Time said it was 1:56, so a little different, but not dramatically (although my wake-up time varied by about 15 minutes between the three devices). Sleep as Android's 'wake detection' is only in beta, and indeed when tracking from the Fossil Marshal it thought I fell asleep a fair bit sooner than I did. The bigger problem I found with this and Sleep++ is that these devices and apps needed me to tell them when I was going to sleep and waking up. Zeitzer told me about a field of cognitive behaviour therapy for insomnia where the focus is not to get people to sleep more, but to convince them not to care so much about their sleep. "If you can convince people not to care so much about it, that's improving their subjective sleep," he said,"which at the end is what we actually want".

Zeitzer talked about how obsessing over sleep data can end up having a negative impact, and I think that's important here, because having to tell some devices when I was going to bed and waking up definitely changed my mentality. Perhaps coincidentally, these wearables picked up far more fluctuations than the other two, and were clearly very sensitive to my movement. Again, this was giving me more to think about, and a lot of this seems to be "noise" that isn't all that useful. Sleep cycles tend to last 90 minutes, and so a little movement between them is to be expected, but with Pebble and Fitbit not picking up so many changes, I'm not too concerned about my overall efficiency.

One thing I did notice is that I'm often awoken when the trackers say I'm in a deeper sleep. Waking up in this state definitely has a negative affect on my overall feeling for quite a while after waking, and it's something I think I can combat by getting an earlier bedtime and perhaps better planning my cycles. So that's one action point already: sleep earlier, plan better.

Zetizer also told me I should take into account variations in my environment and how they might affect my sleep patterns. A good example came on Sunday morning where I was awoken early by someone accidentally buzzing my apartment. While I didn't get out of bed I became fully conscious and moved around a fair bit, and all the devices picked this up.

I'm also going to flag up Thursday here. I've noticed that every morning my sleep starts being more disrupted between 7am and 8am, and this is because of a noise from a generator outside that's disturbing my patterns, even if I don't always become fully conscious. It tends to start whirring at any time between seven and eight and lasts for half an hour. On Thursday it woke me around 20 minutes before my alarm, and both the Pebble and Fitbit Blaze picked up that I was coming out of my deeper sleep at around this time. It's something that's come up during other testing too, and now I'm sure it's something I need to combat, be it by simply sleeping earlier or by blocking it out somehow.

Really, a lot of the nitty gritty details probably aren't worth lingering on right now. This is very surface-level testing for the time being, but even so it has helped me think more about what data is useful, and where I'm going. So, here's what I've learned this week, and my game plan moving forward.

1. Auto-tracking is far more beneficial

Without speaking to accuracy of the tracking once I'm asleep, having to manually tell a device when I'm going to bed or waking up definitely makes me more anxious and aware of how I'm sleeping. To improve my sleep, I feel like I need to be using only devices that auto-track, so that's something to consider.

2. Don't nitpick all the fluctuations

So long as I feel ok the next day and feel like I've got enough sleep - or have an idea of why I haven't - analysing all of the tiny fluctuations in some of my sleep data is probably not going to help me learn anything. On the contrary, it could just make me more anxious. As long as I don't seem to be fully waking up a lot, there's probably no cause for concern.

3. Adjust to outside stimuli

As I mentioned, outside noise is one disruptive element I've observed this week, and particularly troublesome when it's waking me in deeper sleep. To tackle this, I'm going to vaguely plan my cycles so I'm more likely to be awoken in a lighter bit of sleep. Better planning will help me do this. If that fails, earplugs it is.

4. Sleep earlier

I knew this one might be an issue, and although I'm not solely relying on the feedback I've got this week, the information has confirmed that I need to start getting into a better cycle. While I managed earlier bedtimes at the beginning of the week, a combination of events caused my sleep time to be much later in the last few days. One valuable bit of feedback so far is just seeing the times on the screen, and knowing how much I should be improving. Finding the tech that best helps me get to sleep earlier, and more consistently, will clearly be beneficial.

5. Don't oversleep

This goes hand in hand with the above point, and the effects were demonstrated on the Sunday night. I asked Zeitzer about this after looking at the data. "One the one hand, you are probably sleep deprived and need the extra sleep," he said, calling this "the good part" of sleeping in late. "On the other hand, changing you sleep patterns will indirectly, through differences in light exposure, change your circadian clock, which helps your brain to anticipate 'normal' or expected sleep timing. So, this would make falling asleep the next night difficult."

It's clear that I need to create a better cycle for myself, and by going to sleep earlier I hope to do this. Aiming for eight hours seems a good fit for my body, and that's what I'm now trying for, but I might adjust that as time goes on.

Tonight, things get serious. It's the first night I will be tracking my sleep with the Beddit 3, which I'll be using for two weeks. While I'll be able to get information each morning, Beddit is also going to give me an in-depth report analysed by a few experts, so I can put it side by side with other sleep data, in addition to getting useful tips for improving my sleep (hopefully).

Previously

Sleep diary: Week 1 - Separating science from nonsense