[personaldev] Initial review of Zeo home sleep monitoring (EEG) system
Zeo uses an EEG headband to read your brainwave state and determine whether you are awake, dreaming, or in light or deep sleep.
The device is perfect for informavores and quantified self types -- in addition to a holistic sleep quality score, it produces a little graph showing your minute-by-minute sleep activity. It encourages you to experiment with your daily activities to see how they affect your sleep. Given that most people are not self-starters or sustainers of this kind of experimentation, they have created a coaching system that gives you experiments to try and helps you interpret the results.
I've had a professional sleep study done. These studies are incredibly sophisticated and measure a wide variety of things, but have some serious drawbacks. First, you are so covered in sensors that it's really hard to sleep naturally. Second, sleep studies are so expensive that you can only really do them for one or two nights. This makes it very hard to see how various factors affect your sleep.
In contrast, the Zeo is minimally invasive and leaves me full freedom of movement. I can use it as often as I want, allowing me to test my sleep performance under different circumstances.
The downside of the Zeo is that it measures far fewer things, and potentially at a lower quality, than a professional sleep study. It only measures EEG, and only does so in 2-minute increments. This means that short awakenings can be missed entirely. Thus it does not catch apnea unless the apnea is severe enough to cause long awakenings.
Here's what I've learned so far: (some background on sleep stages may be useful if you don't know them)
- My professional sleep study claimed I did not get any deep (stage 3/4) sleep. However, the Zeo claims I get a good amount of deep sleep every night. It's unclear whether this is because the sleep study was on a "bad night" or because the Zeo and the sleep study technicians score sleep differently. Apparently I have high delta brainwave activity in general (even when awake), and it's possible that this is biasing the Zeo. Or perhaps my sleep habits have changed in the four years since I did the sleep study.
- I apparently dream *a lot*. Normal people get 90-120 minutes of REM (dreaming) sleep per night. I get 160-180 minutes. I'm inclined to believe the accuracy of Zeo on REM sleep as its brainwave signature is very distinctive. My sleep study only showed about 60 minutes of REM. Again, it might have been a change in sleep habits over time, or interference due to the sleep study environment.
- As I've long suspected, having intense, emotionally difficult conversations or arguments late at night dramatically negatively impacts my sleep. I only have one data point on this, and I think I'll stop short of encouraging people to call me late at night with stressful things to discuss in order to get more data. :-) On this night I dreamed a lot less and took much longer to get to sleep.
- I tend to not have regular sleep periods. Most of my dreaming occurs in a huge block up to two hours long around 5-7am. I have no idea what this means.
- Zeo claims I wake up 0-1 times per night. My sleep study showed 20 awakenings, most of which were under a minute. Zeo is probably not time-sensitive enough to measure these short awakenings. I feel I wake up more than once per night -- in fact, I'm making a point of looking at the Zeo when I wake up so that I can see how accurate it is. This has the unfortunate side effect of producing dreams in which I look at the Zeo and try to remember the time. I know this is happening because the time on the clock I remember from the dream was later than when I woke up for real.
Some experiments I'd like to run:
- See how periods of intense learning affect my sleep. There are different types of learning of course -- going to a party and meeting lots of new people is very different from spending a day learning to snowboard.
- See what kind of correlation I get between exercise and sleep. I'd vary the amount, time, and type of exercise to see its effects.
- See how consistent vs inconsistent bedtimes affect sleep quality. I know that consistent bedtimes are better than inconsistent ones, but it's a question of how much better. I often sacrifice sleep in the name of fun, and I want to know how much it's costing me, quantitatively speaking.
- See if taking allergy meds at night would improve sleep quality.
Some ways I'd love to see Zeo improved:
- Add a microphone to measure snoring and environmental noise. Snoring tends to have a distinctive enough rhythm that it should be easy to distinguish it from most other types of background noise. This is the sort of thing that Fourier transforms were born to do. :-) Since snoring is indicative of sleep hypopneas or apneas, it's worth measuring. Some apneas are silent though. The microphone could even be on the headband so that it could use bone conduction techniques (like wireless headsets do) in order to isolate noises you make from other noises in the environment. Alternately, an accelerometer could be used to pick up on the physical vibrations of your head as you snore.
- Add a Bluetooth accessory that acts as a lucid dreaming tool. The tool could determine when you're in REM sleep and start giving you audio and visual hints to begin lucid dreaming. The system could even adjust its strength by turning its volume / brightness down if you start waking up.
- Add a photosensor that tracks light levels in the room so that its effect on sleep can be automatically measured.
- Use the EEG headband to monitor concentration and focus levels while awake via alpha and beta brainwave levels. Of course, only people without a shred of self-consciousness would wear it outside their home, but I imagine there are a lot of interesting things that could be measured at home in the evening. For example, my concentration and focus could be monitored during the last hour or two before bed. The device could recommend the optimal time to go to sleep. Earlier in the day, the Zeo could be used as a neurofeedback tool in order to train me to maximize concentration and focus.
- Do whatever it takes to get the measurement sampling period down to a few seconds while preserving accuracy. This would let the device pick up on the frequent short awakenings (microarousals) that chop up sleep and are often correlated with apneas. I think it's somewhat disingenuous to claim, as the Zeo blog did, that this would "overwhelm users with data".
- Open up the system to 3rd party developers so that new peripherals can be created (like the ones mentioned above) and the sleep data can be integrated into other programs that are focused on exercise, weight loss etc.