CPAPtalk.com

gvz,
Pretty scary handling of data for a medical device, don't you think?

I have the CMS-50F. For someone who is even mildly computer literate the set up of the program isn't too bad--that is with the help of the manual clarification from Cooper Medical.

As for fit, it seems to work fairly well for me. The good news and something to consider with other units is that the 50F has a rechargeable battery--with an estimated life of 3 years. The more professional model used regular batteries that I think lasted a couple of days. I don't know if rechargeables would work or not, so you might have to factor in battery cost if you plan to use this daily.

One of the big advantages of the CMS monitors is the program that gvs is developing to be able to up upload the data in the Rescan 3.11 software that works with the newer ResMed units. It is very cool to have the SpO2 data load up with the S9s data upload.

I do not know the accuracy of the SpO2 monitoring aspect of the 50F. I certainly am hopeful that it is accurate. It often seems to correlate with the other data in ResScan 3.11

I have a fairly unique situation so this may not apply to everyone. My heart rate is very irregular and I often have a fast pulse rate (60-160 bpm). Most exercise chest strap monitors can't keep up with the variability of my HR. The CMS 50F also cannot follow my HR accurately, either on the high end or the low end. I know this for a fact as I have a EKG at home and have tracked both at the same time. For someone with a more consistent HR (most people) it might be fine. But if someone has an irregular rate, this monitor will not record that accurately. It is possible that I have a defective unit, but it seems to work well in all other aspects.

Does anyone know if in a situation like mine where the pulse rate is off, could the SpO2 data still be accurate?

frh wrote:I bought a CMS 50D+ from Cooper Medical (he participates here) a couple of months ago. Taping it to my finger overnight was not comfortable for me.

I just had surgery and wore an oximeter for 4 days in the hospital without any discomfort. Pictures of the rubber finger thing on the CMS 50F look like it would be a lot more comfortable to wear than a little plastic brick.

It's hard for me to justify spending money to replace something I just bought, especially since my business has been so slow this year.

I ended up purchasing the 50F because of the soft finger probe. It IS soft, but if it makes you feel any better, it's pretty tight on my fingers and I frequently have finger pain from it in the morning. And my fingers are not unusually fat or thick.

But I still like it and have found that my results are in line with the test results from my DME's monitor.

The BIG problem I see, after thinking about the details of the identified flaw (at least in the CMS50-F): It generally works, but may be way off in areas (abnormal events) for which you bought it in the first place.

* Delete message
* Quote message

Re: Would be interested in your thoughts. thanks

Sent at: Wed Sep 08, 2010 8:00 am
From: Arizona-Willie
To: gpk111
Although I have both the 50-C and the 50-F I have been less than thrilled with them.

I have long suspected their accuracy ... mainly because of the blank lines I would get sometimes. I didn't know what caused them until reading these posts.

I thought the report generated by the other oximeter ( the VirtuOx report ? ) was more informative because it actually tells you how many minutes you spent below 88 % etc. rather than just a percentage.

At one time I developed a spreadsheet where I input the time used from the Rescan and the percentages from the SO2 report and calculated the time at each level. But that's a hassle to mess with all the time.

From what that one person says, there doesn't seem to be ANY reliable recording oximeters. They all give crazy readings from time to time. < shrug >

Looks like the Contec only has 4 second sampling, and doesn't say so anywhere.

The SPO Medical 7500 can let you set the sampling rate at 1.

And the professional, very very expnesive onse sample at mili or microseconds.

O.

Maybe my brain isn't working too well, but what i don't understand is how you can make use of oximeter data, even if the device is accurate. I understand how my Auto M/Encore data readouts can be used to help fine tune the APAP settings, but how do you relate Ox monitor data to anything that gives you a basis for altering your therapy? Or is it just data for the sake of data?

ozij,
The Contec cms50-F appears to sample at 1 second intervals. Reasonable, I think. Didn't know the better units sample more frequently. One of the main topics of discussion (even though the original thread had a slightly different theme) was the accuracy and reliability of the data. To that end, the cms50-F produces a spreadsheet in which the O2 and HR readings are recorded in 1 second increments.

The discussion further speculated about the meaning and treatment of zeros. Zero presumably represents a condition during which the sensor didn't pick a reading. We determined that if there are a lot of zeros (more than say 2 minutes), the resultant report produces a flat line. That seems rational. eg if you remove the probe from your finger, that's rational. However, if there are only a few zeros (say 5 - 20 seconds), it's unclear how those zeros are reflected in the final report. We speculate that they could be
a) ignored
b) averaged in with the rest of the data
c) replaced with the last known "real" data

I'm currently doing additional investigations to see how zeros are treated. The software could be screening them out or they may lead to erroneous interpretations.

Sleep_depraved:

Since one of the major detrimental side effects of OAS is the reduction of oxygen, an oximeter measures that effect directly. The tracking on the xPAP machines only measures the breathing patterns, not their effects. If one has invested enough time and money, a correlation of sleep events and O2 desaturations (along with heart rates) can be correlated to give a more accurate picture of your sleep patterns.

The concern is that the resultant treatment of zeros could be interpreted and reported as an "Event." The definition of an event can usually be set at different levels in the software. For example, mine is set as follows: For O2, it's a drop of 4% in saturation for at least 10 seconds. For HR, an event is defined as a change in rate in either direction of at least 6 bpm for a minimum of 8 seconds.

I analyzed about 6 1/2 hours of raw data recorded by the Contec CMS50-F wrist oximeter (it's captured in 1 second increments) and compared that to the report generated by their SpO2 software.

DATA CAPTURE
I used some band aids to secure the plastic sensor to get as high a proportion of valid readings as possible. Data was recorded from about 1 AM to 8 AM. I slept about 6 of those 7 hours.

SOURCE and OBSERVATIONS
The data was taken from the CSV file generated by the SpO2 software. There were about 25,000 observations (7 hours X 60 minutes/hour X 60sec/min) of pulse (heart rate) and SpO2 (O2) levels. The average O2 level for that entire period was 94.2 with a standard deviation of 6.3. The average heartrate was 49.8 with a standard deviation of 5.9.

IDENTIFYING "EVENTS" IN THE RAW DATA
Events are not too difficult to identify. It's easy to see the difference of values using simple arithmetic. I used fuzzy logic (as opposed to rigid rules) to identify a total of 21 events in my direct inspection of the raw data. This compares to a total of 24 events reported by the software with rigid rules.

TYPES OF "EVENTS"
Of the 24 events I identified, I concluded there were two events which were clear aepnea events. The O2 levels went down by about 5 percentage points each time, the heart rate went up by 12 and 20 beats per minute. The events lasted 16 and 20 seconds respectively. In addition,there were 10 "heart rate events" were the heart rate went up by an average of 10 BPM. Those 10 events lasted an average of 22 seconds. The third type of event had to do with a lack of reading, or zeros in the data (zero events). There were a total of 9 such occurrences. In each case zeros were recorded for both the O2 and the heart rate, clearly pointing to a lack of data, rather than an organic malfunction. The zeros were recorded an average of 8 seconds with a range of 5 to 11. The events triggered by those zeros, however, could last much longer, since it would include the time for the tracking algorithm to return to "normal." In one case, however, the data picked recprded immediately after the zeros was the same as before the zeros were recorded.

COMPARISON TO REPORTED EVENTS
The SpO2 software reports both SpO2 events and Pulse (Heart Rate) events. For this same period, it reported 4 O2 events and 20 heart rate events. This contrasts to the 2 O2 (aepnea) events and 10 pulse (heart rate) events observed in the raw data. The raw data surrounding the "zero events" led me to the conclusion that those events should not be counted as events at all (neither an O2 or Pulse event). In most of those 9 "null" cases, the conclusion was obvious, but in one or two cases, it was possible that a "real" event could have occurred at the same time. The unpredictable time to get back to a "normal" level after the lack of input added to the complexity surrounding the "zero events."

CONCLUSION
Consumer oximeters are surprisingly accurate in ideal conditions, especially given their costs. However, the occasional lack of sensor input in a realworld environment and the incorrect or cavalier handling of that (lack of) data by the software can easily lead to inaccurate summary data and flawed conclusions. These comments apply to the CMS50-F and its associated software package, SpO2 v0.981.

I'd be happy to share the raw data and analysis spreadsheets.

Sleep_Depraved wrote:Maybe my brain isn't working too well, but what i don't understand is how you can make use of oximeter data, even if the device is accurate. I understand how my Auto M/Encore data readouts can be used to help fine tune the APAP settings, but how do you relate Ox monitor data to anything that gives you a basis for altering your therapy? Or is it just data for the sake of data?

Here's how I use the oxy data:

1. If an event that appears on the Encore readout (OA, CA, etc) is accompanied by a degree of desat, I assume the event on the Encore printout is valid. But some of the events (I'm speaking of my printouts), especially CA's, are not accompanied by a change in O2. I suspect these "events" reported on the printout are either trivial or spurious.

2. The AHI is the number of events per hour, but does not measure their severity. An acceptable AHI of, say, 3.5 may still be accompanied be significant periods of desaturation below 90%. That would call for some change in the XPAP settings. OTOH, a the same AHI consisting of a lot of CA's without desaturation, is probably not worth an intervention.

3. If the O2 saturation remains high and the AHI low then they confirm each other, and I'm reassured that the XPAP and its settings are working as they should.

M.D.Hosehead wrote: 2. The AHI is the number of events per hour, but does not measure their severity. An acceptable AHI of, say, 3.5 may still be accompanied be significant periods of desaturation below 90%. That would call for some change in the XPAP settings. OTOH, a the same AHI consisting of a lot of CA's without desaturation, is probably not worth an intervention.

3. If the O2 saturation remains high and the AHI low then they confirm each other, and I'm reassured that the XPAP and its settings are working as they should.

With Encore Pro, and James program Encore Analyzer, and you have a way to read how long the destat lasted, very handy. However the latest and greatest machines (PR1's) require software that can't use Encore Analyzer Jim

Based upon some of the mention of flat lines indicating a problem with sampling of the SpO2, does my graph look like a sampling problem which would make the data questionable?

When I wear the monitor it seems to respond to me being quiet by usually showing a decrease and when more active by showing in increase. So I have been hoping the SpO2 portion was accurate. But how to know?

(I am the person with the arrhythmia who has determined via my EKG that the pulse rate data is not accurate.) But my oxygenation has generally seemed pretty good--at least on my 2 sleep studies. Every once in awhile the 50F will report a very steep but brief drop off of SpO2 (not pictured), but then quickly recovers. I have thought that these infrequent drop off are simply that I moved during the night and temporarily displaced the monitor.

Most of my night goes similar to the below so I'd consider it representative--certainly no jagged line tracings for me.

gvz wrote:From my previous experiment, if the flat lines aren't colored in grey, then you should be fine.

Cool. I saw the grey line but thought you had added that so the post would show up better. Do you know if it is typical that there would be so much flat-lining throughout the night?

ldj,
1. On the surface, your graph looks very credible for the hour you show.
2. How do you attach a picture to the post? I have yet to learn that.
3. I actually took my pulse and determined that the pulse data is credible (at least on the CMS50-F wrist model).
4. The O2 graph looks fine. Without looking at the raw data (1 second samples), I would say the flat lines just mean that the O2 readings are the same from one reading to the next. I would also interpret it as getting real and continuous readings (as opposed to skipped readings - see #5 below).

gvz and ldj,
5. From my research with raw data, if there were misreadings (either "bad" readings - or more probably - zeros), there would be disruptions in the green line (as opposed to a flat green line). A grey line only appears when there are a lot of zeros (eg you take off the sensor, but don't turn off "record"). I don't know exactly how many seconds of zeros it needs to show a grey line. I'm guessing it has something to do with the scale of the graph. In my case (less than 20 zeros shown in a 6 hour graph), that anomoly always showed up as a glitch in the graph.

Hope that helps

gpk111 wrote:ldj,
1. On the surface, your graph looks very credible for the hour you show.
2. How do you attach a picture to the post? I have yet to learn that.
3. I actually took my pulse and determined that the pulse data is credible (at least on the CMS50-F wrist model).
4. The O2 graph looks fine. Without looking at the raw data (1 second samples), I would say the flat lines just mean that the O2 readings are the same from one reading to the next. I would also interpret it as getting real and continuous readings (as opposed to skipped readings - see #5 below).

gvz and ldj,
5. From my research with raw data, if there were misreadings (either "bad" readings - or more probably - zeros), there would be disruptions in the green line (as opposed to a flat green line). A grey line only appears when there are a lot of zeros (eg you take off the sensor, but don't turn off "record"). I don't know exactly how many seconds of zeros it needs to show a grey line. I'm guessing it has something to do with the scale of the graph. In my case (less than 20 zeros shown in a 6 hour graph), that anomoly always showed up as a glitch in the graph.

Hope that helps

gpk111,

Thanks for the response. Maybe I can make the 50F to work for me. It would be nice to have a more accurate pulse rate reading, but with my situation I can live with it if the SpO2 is accurate. I really don't know if a more professional monitor would do any better with the pulse.

As for the graphics, I capture the screen shot with a great program called SnagIt (highly recommended). But I think other members capture with printscreen and then put it into a graphics program for editing and exporting, I then save it as a jpg and upload it to Photobucket. From Photobucket I copy the "IMG" address and paste the link into my post where I want the graphic to appear. Then check the preview button in the post editor to see that everything looks OK before I submit the post.