Category Archives: Sleep

Device/software designed to analyse sleep pattterns

First Thoughts on Beddit: passive biometric sleep monitoring

I’ve written many times in the past about passive, automatic non-invasive life-logging and sleep is the one area my participants (and I) have been constantly excited about and disappointed.

Sleep Cycle stats on iPhoneThe original popular player in the field, the iPhone SleepCycle app, had great promise and gave some indication of sleep quality for people who slept in a certain way and had a certain type of mattress. As I noted in my review, it fell down in terms of accuracy for some people and in its ability show export its data so I could use it elsewhere. I reviewed and tested many of the wrist actigraphy devices, including putting the fitbit into its bracelet, but it suffered from even more wild accuracy problems (sound sleepers were told they woke up 40 times in the night) and it was quite annoying to use.

ZeoOnHeadCertainly the gold standard of consumer sleep monitoring was the Zeo which couldn’t be fooled into believing you were in deep sleep by lying perfectly still. However it was a bit invasive (to say the least) and although it began to allow data exporting for analysis in the months before the company went bust, most participants reported that the data was interesting but the effort involved exceeded the value they got from it in many cases (not to mention the fact that they felt silly wearing the headband.

So I was intrigued, if a bit sceptical, when I saw the indiegogo campaign for a passive sleep monitor that actually measured any biometric activity. Gear4 Renew Sleep Clock I had tried one other passive monitor, the Renew Sleep Clock, but was disappointed in that in was about as inaccurate as the wrist based accelerometers and also had poor data export capabilities. It relied on ultrasonics to measure movement of the space in front of the clock.

Beddit sleep monitor installed on mattress

 

When my Beddit sleep monitor arrived yesterday I was still a bit sceptical. Visually, it is a thin plastic strip that goes on the mattress (in my case a mattress topper) just under the sheet with the control module hanging over the side of the bed. It stays permanently plugged into 900 mA USB power supply which is supplied. The free iPhone app (Android app is apparently on the way but not quite there yet) has very basic controls. You can optionally enter some physical characteristics and optionally set an alarm time (which I did) but other than that there is just a start sleeping button which starts recording data. Data recording stops when you stop the alarm but presumably you can manually switch off if you don’t use the alarm (will try that tonight). This first version of the app has no data export facility other than the ability to post your sleep score to Facebook or Twitter with some pre-formatted text giving your raw total sleep score, which I’m not complaining about as I know that I’m effectively a post production beta tester. Therefore to show you the results I’ve taken screen shots as I scrolled down the one page of output.

Beddit app output Night 1 page 1 of 3On the top of the screen you can see my overall sleep score. I’m not sure what the maximum is or what the scales are but it looks like a percentage and the “coach” suggests that I could do better! I set the sleep goal in the preferences to 8h but went to bed very late so I below points there obviously, getting only 67 points because that is 67% of my goal (5h23m out of 8h). I guess I could get over 100% by setting a lower sleep goal. I didn’t wake up or get out of bed so I get bonus points there (do they go negative if I do? must check) and my wife will be glad to know that I didn’t snore, although she is away at the moment so I don’t have any independent evidence. Tonight I’ll cross check with my Sleep as Android app which records an amazing set of data and produces lovely graphs of it, including sleep talking and snoring.

Beddit Night 1 page 2 of 3Scrolling down I could open the generic sleep advice but you can get that anywhere, I don’t think it is very tailored to the data yet. Next we see the sleep total which is where most of my score comes from and my resting heart rate which I was surprised it could do through the sheets.  I know from all my other measurements that this is indeed my resting heart rate so I’m already impressed! How this translates to sleep quality measurement I will be very interested to see.

Beddit Night 1 page 3 of 3It seems to have got my time to fall asleep about right (so it beats the apps that sit on the mattress on that score) but other than that the measure of sleep quality remains to be seen. From a convenience point of view it beats the phone apps because it isn’t going to fall off the bed which was the biggest complaint from my particpants and if it records only one person when there are two in the bed that will be an advantage. My wife and I have very different resting heart rates so it will be easy to tell if it is measuring me, so as long as I can remember whether or not I was awake when it says I was then I should be able to check that.  So it is a qualified thumbs up from me, but I’m looking forward to being able to get the data out of it so I can do my own analysis.

 

How do you sleep the night before you know you are going to be live on national TV?

On Monday I got my 15 minutes of fame all in one day. Celia and I were on the BBC 1 Breakfast TV sofa for about 6 minutes…

BreakfastTV

 

 

 

 

 

 

 

 

 

 

 

Then I did a Radio 4 interview where I mentioned sex, excrement and spying litter bins all before 8:30 in the morning….followed by BBC Radio Scotland then BBC Radio Warwickshire the next day.

Celia, who was my premier league lifelogger from the study, thought we should monitor our sleep for the night before we knew we would both be live on national TV, so we wore our Zeo headbands. Now I am a championship sleeper, when WakeMate were still in business and I used their wristband (the most comfortable of the lot I might add), I was regularly in the top 5 ranked sleepers in the world on their charts.  However, it appears that, at least for me, knowing your are going to be live on national TV affects sleep. Here is my Zeo sleep graph for that night:

Blaine's sleep the night before Breakfast TV Broadcast

As you can see, I take ages to fall asleep, have very little deep sleep (dark green) and very little REM sleep (light green), then wide awake from 4AM wondering which talking point I can use! Finally around 6AM I fall back into REM sleep and the alarm goes off.

Meanwhile, here is Celia’s graph (different format because it came from an Android phone):

CeliaNightBeforeBreakfastTV

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

As you can see, she sleeps right through the night (other than the 2 wake up spikes which are normal for her), she gets up at 6 then takes the headband off later. Celia also gave a brilliant interview, so from the small sample we conclude that, for some people, the threat of appearing on live television has an adverse affect on sleep quality, while for others, it has no effect. Another data point for the journal of useless results….

A new activity tracker which also includes heart rate, but not automated…

Withings have announced their competition for the FitBit, the Pulse. It competes directly with the FitBit One but additionally tracks heart rate if you put your finger on the sensor (not automatically as we would have hoped, for that you need the Basis). I still use the FitBit Zip in preference to all the others because it overcomes one of the major usability faults my participants have complained of: the charge sync cycle. The Zip will store its data for up to a week but will opportunistically sync whenever it finds any fitbit low power bluetooth station, including a smartphone app. I don’t have to do anything and I get an email if my zip hasn’t found a fitbit station within a few days.  The FitBit One needs to be charged at least once a week while the new Pulse claims a 2 week charge life. If that’s true then for me it would probably be good enough to switch from the Zip (battery replacement every 2-3 months).  Although Maarten suggest the HR is not very accurate (as some have claimed about Basis):

 

@yankeeinlondon bulky, great display, inaccurate/slow heartbeat measurements, nice app integration

— Maarten den Braber (@mdbraber) June 23, 2013

Actigraphy

Actigraphy is a simpler alternative to expensive laboratory techniques for measuring how deeply a person is sleeping. It is explained in a paper by Ancoli-Israel et al. (2003) in the journal SLEEP. It involves measuring how still a person is while he or she is sleeping with complete stillness equating to deep sleep. Typically an accelerometer is attached to the wrist of the non-dominant hand and movement over time is recorded and is the basis for a number of consumer sleep monitoring devices (e.g. FitBit, WakeMake) and mobile phone apps (e.g. SleepCycle, Sleep as Android),

WakeMate Sleep monitor wristband from Wakelytics

[UPDATE: WakeMate is no longer being sold]

WakeMate measures the quality of a person’s sleep based on actigraphy. It consists of a small accelerometer placed inside a soft elasticated wristband. The device is first charged for a few hours usings a mini-USB adapter. One charge lasts for about two nights of use if the device is switched off between uses. Before going to sleep, the user activates a small switch inside the unit tucked into the wristband and an LED appears. Then a custom app is activated–the device is factory set to work with only one mobile architecture, either Apple iOS, Android, or Blackberry. The user then makes sure the app has a bluetooth connection to the wristband and sets the desired wake time. As with other sleep monitoring devices, the software will sound an alarm in advance of the desired wake time if it detects the user is in a light sleep phase so that they are not accidentally woken from a deep sleep leaving them more groggy upon waking.

Ease of Use

The device and software are somewhat complicated to use because of the need to switch on and off, charge often, and the multiple button presses to activate the software and ensure a bluetooth connection is active.

Cost

This is one of the least expensive lifelogging devices available at time of writing at approximately US$60.

Interoperability

Although WakeMate is subscription free and provides unlimited access to user data on the website (or the app), there is no export function and I have found no way to scape data from the web interface, although their Wakelytics website does have the most comprehensive analysis of the actigraphy data that I have seen compared to other devices. They also allow comparison with the whole Wakemate user base. When I use the device, I often appear in the top 10 and will therefore campaign for sleeping to be entered as an Olympic sport.

Here is a sample of the within user comparison bar chart showing sleep pattern over successive days:


Day by Day comparison.

Here is a comparison of user data over the night, lifetime, and compared with the whole user base:

Global Comparison

This graph shows more detail of the classification of the sleep over an individual night:

Night Classification

And finally, the raw movement data:

The WakeMate is less intrusive than the FitBit wrist strap but more tricky to operature as you have to flick a small microswitch, then fire up an app on your phone, make sure Bluetooth is on, then press a button at the right time and the battery only lasts about 2 nights so there is a lot of re-charging involved. It does produce more fine-grained data than the FitBit and is less expensive, but the electronics seem a little unreliable, as mine stopped working after a few months. At time of writing there was no data export facility.

Sleep Cycle

The Sleep Cycle iPhone app is probably the most popular (being one of the earliest) and is fairly automatic, it is simply switched on each night with the charger plugged in and placed on the bed face down beside your pillow. It measures movement and sound and wakes you within half an hour of your chosen wake up time by looking at how active/noisy you are and determining when you are moving from deep sleep to light sleep. The idea is that it wakes you when you are entering light sleep so you don’t feel groggy in the morning. You can also set it to just record your sleep levels. Either way it produces a page of simple stats with a graph:

While easy to use and automatic, the data produced is only available in the form above. It can be e-mailed as an image or posted on a social network, but the underlying numerical data is not available. Like other smartphone apps using this method, it isn’t true actigraphy and results are highly dependent on the mattress type and whether or not the person sleeps alone.

Zeo Personal Sleep Manager

The Zeo Personal Sleep Manager is a mains powered bedside alarmclock that records the most comprehensive data on sleep activity of any of the sleep monitoring devices I have tested. In addition to being the most expensive, it is also one of the more intrusive in that it requires the user to wear an elasticated headband while sleeping:

While the other sleep monitoring devices rely on the relatively imprecise science of actigraphy, this device uses proprietary technology to apparently monitor brain activity using a simplified form of electroencephagraphy, although this is only a guess. I have attempted to determine if the device is simply measuring heart rate, movement or some other metric by wearing it on my arm instead of my forehead but it didn’t record any data so I have no reason to doubt that it is measuring some kind of brain activity.

The main part of the system is the bedside digital clock which also stores and charges the headband when not in use:

In common with the other sleep logging systems, the clock allows you to set a desired wake time and will wake you before the wake time if you enter a light sleep phase in order to prevent you having to be awkened (groggy) from a deep sleep phase later. The data is stored on an SD card. In order to get at the data you need to remove the card, connect it to a computer, log into the zeo website, then upload the data from the card. Here is a basic view from the website:

Initial Zeo Sleep Data Screen

and you can examine an individual night to look at the apparent level of sleep (based on brain activity?) so you can see how long you were in each of the four stages of sleep:

Zeo Detailed data for a given night

 

There are now smaller versions which connect to  a smartphone app instead of using a less portable alarm clock, but at between $100 and $200 this is an expensive solution and some users find the headband uncomfortable or difficult to get to sleep with (besides looking a little silly).

On the plus side, the manufacturer provides free unlimited access to your own data on their website after you have bought the product and they allow you to export your own data in an open format, so once you have bought the device there are no other costs. On the downside the device isn’t Internet enabled so you have to manually move the SD card and go through an upload process when you want to get at your data. The newer mobile version which connects to a smartphone doesn’t appear to have this limitation but I haven’t evaluated it. I’m also impressed with the way the company seems commited to keeping all the data open and letting people hack it. Brian Schiffler hacked the Zeo to produce real time brain activity data and Zeo hired him as an intern to make the data more open.