Zeo Personal Sleep Coach?

[robertjvan]

Just seen an AD on TV, so maybe it's good.

[rada]

I have been studying my sleep with ResScan data, pulse oximeter, and webcam data. I am beginning to get a handle on the relationship between respirator, SpO2, AHI, and my position in bed. However, I continue to be haunted by the feeling that something is missing and that an electroencephalogram (EEG) might be needed? What I mean by haunting is that I feel something about my quality of sleep is missing with the respirator sleep. For instance, today I slept for 6 hours with the respirator and my SpO2 was good. After those 6 hours I could not sleep with the respirator anymore and was deciding to rise to begin the day's chores. However, somehow I felt dissatisfied and decided to try sleeping without the respirator. I slept another two hours without the respirator. I have experienced this kind of thing many times. Was that non-respirator sleep putting me into some phase of REM or NREM sleep that is important to health? I can guess my sleep stages from the knowledge of normal sleep rhythms and my current data, but I also wonder whether something like an EEG would help. I was delighted to find this morning on cpaptalk.com a long thread about Zeo Personal Sleep Coach. Someone also had spoken about this haunting sensation that some aspect of brain rest was disturbed with the respirator. The jury seems to remain in deliberation on the value of this Zeo device for people with OSA, but I intend to investigate further and either acquire a Zeo or something else that tracks brain sleep stages. If anyone has further suggestions in this regard at this stage, then I would welcome them.

[BlueHairBob]

I use a Zeo and APAP each night. This allows me to correlate my breathing and my sleep phase data. This has helped me make more progress in the past few weeks than I have been able to in the past few days. The Zeo does report me being in a light sleep at time when I could swear that I was awake, but that is probably because I do go into a very very light doze at times. I am so glad for the data I get from the Zeo.

[rada]

Bob, Thank you for the feedback. Do you feel that the Zeo is measuring brain status or something else? Do you wonder whether an EEG would add much value? I take your feedback as a big plus for Zeo, as I continue down this path of trying to get to grips with my situation. You said weeks versus days when I guess you meant weeks versus years - -is that right?
Yours, Roy

[BlueHairBob]

Yes, I meant weeks vs. years. Sorry about that.

The way I understand it, Zeo uses different technology than an EEG. An EEG may be more accurate, but what Zeo does is helpful. It appears that Zeo takes the sensor data and feeds in into a neural network that decides what stage of sleep is represented. Neural networks are interesting: you feed them data and then tell them what to think of it; then, after a while, the neural network takes data and tells you what it thinks of it. In this case, it would be like getting a reading from the Zeo sensors (without interpreting it at all) and, using something like an EEG, telling the Zeo "This person is currently in REM". At another point, "This person is awake", etc. Eventually, the Zeo neural network takes raw data and says "This person is in deep sleep." You give it two inputs: one is the sensor data you don't know how to interpret and the other is data from a different input source you do know how to interpret.

Neural networks are a funny area of computer science in that you don't have to program an algorithm exactly - it is more like you just give the computer samples and tell it what they mean, and the neural network learns the pattern recognition. My understanding is that is basically how Zero works. Based on some (very limited) work I have done with neural networks in the past, it makes perfect sense to me that Zeo can work.

An EEG may be more accurate, but I really like how easy the Zeo makes it to get some "interpretable" results.

[rada]

Bob, Thanks for the explanation. I have now searched on google scholar and National Library of Medicine's pubmed for anything about Zeo without success. I imagine that I could use existing data such as pulse, activity in bed, and time asleep to try to forecast sleep stage and might be able to train a neural network to roughly predict a person's sleep stage based on such data. What is unclear to me is the extent to which the '3 sensors' of Zeo are reading brain waves and making a sleep stage assessment as defined by EEG patterns. I understand that for you it helps and that certainly is what matters. Thanks again, Roy

[jskinner]

The jury seems to remain in deliberation on the value of this Zeo device for people with OSA, but I intend to investigate further and either acquire a Zeo or something else that tracks brain sleep stages.

To the best of my knowledge Zeo is currently the only Consumer level sleep stage monitor on the market.

What is unclear to me is the extent to which the '3 sensors' of Zeo are reading brain waves and making a sleep stage assessment as defined by EEG patterns.

Are you suggesting that the company is lying about how the Zeo works? Its a 3 channel EEG with automated scoring. I don't understand the high level of skepticism about the Zeo here. The only thing to question is how accurate the automated scoring is.

I've been using my Zeo nightly since September 2009. I'm happy with it.

[BlueHairBob]

jskinner, it was the fact that you use one that convinced me to get one. I am very happy I did. Thanks for you earlier post showing the graphs produced by your unit. Bob

[rada]

Hi James,
I based my concern on the comments earlier in this thread. Yellow asked this: "The literature claims it measures your brainwaves. Does it really do that (and for all phases of sleep) or is that just a figure of speech? Is it instead that it detects from slight movements of the skin on your forehead when you eyes are moving (and calls it REM), and perhaps it also has a traditional motion detector that tells it when you're walking around or tossing around (and calls it Wake)." Derek from Zeo said: "It’s not just a figure of speech - Zeo does indeed track your brainwaves in order to determine sleep phases. Your questions are getting at another component that we don’t mention in the literature though - other factors that create electrical signals, like body and eye movement, are also taken into consideration. After all of these inputs, the rules that help Zeo decide which phase of sleep you are in are quite complex." How many sensors for brain waves does Zeo have? A measure of brain sleep activity would seem to require multiple sensors for brain waves.
Zeo cites these publications: "Wright K, Johnstone J, Fabregas SE, Shambroom JR. Evaluation of a portable dry sensor-based automatic sleep monitoring system. Sleep. 2008;31 (Suppl.):A337, Shambroom JR, Johnstone J, Fabregas SE. Evaluation of a portable monitor for staging sleep. Sleep. 2009;32 (Suppl.):A386)." However, I cannot find evidence for these as publications of full papers but rather that they were presentations at a conference and thus generally not archived nor accessible to us, whereas a bona fide journal publication would be archived in publically accessible libraries.
Given that Zeo uses a neural network, then the inputs are the sensor data and the output is the stage of sleep. What is the training function? Is the training done one time on data at headquarters that produces an algorithm incorporated in all products or is the training done continuously for each device with its users?
I am keen on tracking brain activity in my seemingly never-ending effort to come to peace with my OSA treatment. I now use AHI, pulse-oximeter, and webcam but feel that a better handle on the brain activity would be useful. I worked 35 years ago with a lab that outfitted EEG head caps for astronauts that were highly portable but fully functional as EEG devices and suspect that something like that remains available somewhere. I do medical and artificial intelligence research and could build my own machine learning program that took my ResScan data, my CMS50E data, my coded webcam data, libraries of correlations among various time series and sleep stages, and produce a program that tried to stage my sleep. If I buy another recording device in order to get to brain status during sleep, then I want to be distinguishing from the other data I already have and the brain wave data.
Zeo's product would seem to be more convenient to acquire and use than the currently available, commercial alternatives for the public. How can I learn more about Zeo's input and how it converts that to output?
I am impressed with your testimonial for Zeo.
Thank you,
Roy

[jskinner]

How many sensors for brain waves does Zeo have?

It has three EEG channels.

A measure of brain sleep activity would seem to require multiple sensors for brain waves.

Not so. Philp Low has developed a single channel sleep stage algorithm called SPEARS. His company, NeuroVigil, is working on bringing a medical quality sleep stage device to market based on this impressive technology.

How can I learn more about Zeo's input and how it converts that to output?

Here is a high level explanation of how Zeo works
http://blog.myzeo.com/5-steps-to-phasing-sleep/

[rada]

Thank you for the link to the zeo explanation and also to noting that one lead might be enough for some purposes. The Zeo explanation says: "The Zeo headband uses 3 silverized, conductive, fabric sensors to collect the tiny electrical signals naturally produced by your brain, muscle tone and eye movement." You said it has 3 EEG leads. If it has exactly 3 sensors and collects signals from brain, muscle tone, and eye movement, then does that mean that some sensors are simultaneously acting as EEG leads and as muscle sensors?

[BlueHairBob]

I had previously sent email to the CTO of Zeo about some items of interest and he wrote back within a few hours. His email address is published in the Zeo forums.

[el_zorro]

I bought one of these a few weeks ago on the Amazon deal and have really been impressed with it. It works well and have been used with a CPAP nasal mask with no problems. If you are used to a mask, you will hard even notice that you are wearing the small, light, comfortable headband. The part that is most interesting part for me was that I seem to go into deep sleep earlier on in the night and then typically go in and out of REM sleep for the remainder of the night. It is very accurate in capturing the wake cycle, there was one night that I woke up and couldnt get back to sleep, it recorded this time very accurately.

The biggest advantage for me is that it raises your awareness of your sleep quantity and quality and records it. Just like monitoring AHI, etc, it helps you see where you are at. My AHI numbers are really good these days (1.5 or lower) so I havent tried to correlate the few wake events with breathing events yet but after I capture enough data I will look at the graphs side-by-side.

It would be interesting to see what it does for non-diagnosed OSA patients and if it is a more affordable option to a sleep study for non-complex OSA cases along with an APAP for the titration study. Of course, the sleep lab industry will be opposed to this but anything that would threaten their profit margins I guess would be viewed as a threat.

[roster]

It would be interesting to see what it does for non-diagnosed OSA patients and if it is a more affordable option to a sleep study for non-complex OSA cases along with an APAP for the titration study. Of course, the sleep lab industry will be opposed to this but anything that would threaten their profit margins I guess would be viewed as a threat.

I am interested to hear how it would diagnose OSA. I might buy one to loan out to friends and relatives who are in denial about their apnea symptoms.

But the data would need to be convincing because these people are determined deniers.

[jskinner]

If it has exactly 3 sensors and collects signals from brain, muscle tone, and eye movement, then does that mean that some sensors are simultaneously acting as EEG leads and as muscle sensors?

I guess I should have said it has 3 channels/sensors. Which leads are used for EEG, EMG, EOG are unknown by me.

I had previously sent email to the CTO of Zeo about some items of interest and he wrote back within a few hours.

The CTO has been very responsive to my emails as well. Well he was at least until I suggested they consider using the Neurovigil algorithm in future versions of the Zeo

The biggest advantage for me is that it raises your awareness of your sleep quantity and quality and records it.

Agreed.

It would be interesting to see what it does for non-diagnosed OSA patients and if it is a more affordable option to a sleep study for non-complex OSA cases

Unfortunately, the resolution that Zeo reports for awakenings does not make it useful for OSA detection. Basically it will not report the micro awakeings seen in sleep disordered breathing. This is the biggest drawback of the device in my opinion, as least from an apnea patient point of view. Keep in mind however that the Zeo was never indented as a medical device.

I'm still searching for a more accurate/detialed device but the Zeo is far better than anything we had up until it came on the market.

I am interested to hear how it would diagnose OSA. I might buy one to loan out to friends and relatives who are in denial about their apnea symptoms. But the data would need to be convincing because these people are determined deniers.

At this point I think the Watch-PAT200 is probably one of the easier at home tests to use with 'deniers'