How Accurate is the AHI # on the Respironics LCD?

[cinco777]

When I started reading this forum in March, I ran across a number of threads & postings where members commented on the accuracy/inaccuracy of the Apn-Hyp Index (AHI) numbers (7-day avg and 30-day avg) displayed on the Respironics LCD. Some said they weren't accurate. Others said they were averages of averages, etc. A month ago, I took a cursory look at the numbers and concluded that the LCD AHI numbers displayed on my M-Series AutoCPAP accurately represented the sleep results for my "machine's reporting day". I just completed a more comprehensive 30-day study of the LCD numbers.

Since the machine's internal clock is UTC/GMT, its time will only be the same as your local time if you live near London, England. This is a key fact to remember for anyone monitoring their nightly results using the numbers displayed on their machine's LCD. For example, here on the West Coast, which is now using Pacific Daylight Time, local time is UTC/GMT - 7 hours. Since Respironics machines define a day as running from Noon one day to Noon the next, my machine's Noon-to-Noon day really runs from 5AM local time one day to 5AM local time the next day. When this fact is taken into consideration, I have found that the LCD displayed 7-day avg AHI agrees with the 7-day avg AHI values that can be calculated using the EncorePro database tables (Time-stamped event data in combination with the machine internally collected summary event data). If you are on the West Coast like me and get up about 7AM each morning, you have a slight problem. The machine's night ended at 5AM so you will not see the results for your night's sleep which ended at 7AM local time. You will see your numbers for a sleep night that ended at 5AM local time. Folks on the East Coast, with UTC/GMT - 4 hours, that get up at 7AM local time, will do OK as their machine's day will end at 8AM local time, an hour after they get up. Their machine LCD numbers will include & report their entire night of sleep.

For the 30-day period (4 weeks & 2 days) just Studied, the machine's 7-day average AHI is no different (0%) than the 7-day average AHI (rounded) determined from the data in the EncorePro database. Please note that the LCD 7-day average is a "rolling" 7-day average. That means that it is an average for the past seven days. The LCD also shows a 7-day average for Therapy Usage Hours, System Leak (which is a simple average not a weighted average), and 90% Pressure (not studied). 7-Day Avg Therapy Usage Hours and 7-day Avg System Leak are also accurate within reporting granularities and predicted machine hardware design. Please peruse the attached chart for more details. The 30-day average #s are also accurate within reporting granularities.

[jdm2857]

What a thorough analysis. But at users now know that their machine can do math.

I wonder if the new System One APAP has a clock that can be set. Noon-noon local time makes sense. Noon-noon in one time zone does not.l

[cinco777]

jdm2857 wrote

Noon-noon local time makes sense. Noon-noon in one time zone does not.l

I absolutely agree. Respironics, on their LCD display, provides a user with numbers that only a "geek" may be able to figure out and understand after considerable research and investigation. Using UTC/GMT time internally, without providing a local time reset option, is just one of the Respironics machine reporting problems. It may even be the Biggest "usage" problem for users without the software. Other Big problems include: 1) 7-day average and 30-day average on the LCD and NO one-night report! -- What were they smoking/drinking/thinking? 2) recording/reporting granularities of 30 seconds for the time-stamped events! Boy, that is one way to conceal cause and effect relationships, 3) reporting/recording granularity of 7 LPM for Leaks, 4) only recording/reporting one sleep event (except for VS & VB which may appear in the same interval) in a 30-second reporting time interval! - this leads to missing "tick" marks for "multiples" in EP and user-generated graphical reports that show when events occurred, and 5) restricting user purchase of EncorePro with all its database capabilities that we technical types can use to source all our user generated analyses and reports.

Patient/user friendly was not one of the design priorities for Respironics!

[jdm2857]

I don't think that anyone at Respironics looks at their products from the users perspective. They market them to sleep doctors and DME's and that's who they aim to please. Look at the brochure for the new System One. It's aimed at the home care provider.

I'm not even sure that they realize that their machines are used in rooms where people sleep. If they did, how could they come up with a noisy, whiny machine with too-bright blue LEDs and a high-leak alarm that is too quiet to awaken a sleeping person. (Several have commented that they can't hear the alarm over the sound of their machine and mask.)

Leaves me baffled.

I am, however, very impressed with the testing and patent-reading that you and a couple of others have done into the internal operation of Respironics machines. Patents, as you surely know, are not light reading.

[cinco777]

jdm2857 wrote

I am, however, very impressed with the testing and patent-reading that you and a couple of others have done into the internal operation of Respironics machines. Patents, as you surely know, are not light reading.

I am still a Newbie to CPAP. Re the patent reading, I am following in the footsteps of Muffy, Velbor, SWS, and others who seem to have been studying the patents for quite a while. Since I know spreadsheets (I must have created a 1000 in the last 20 years) and recently developed one to report my CPAP usage, I have decided to focus my efforts on investigating the sleep related numbers that our machines report - pretty much the Where, What, How, Why, Who, etc., - all the questions that a curious techie would want to answer! When I was working (I retired 10 years ago), I sometimes reverse engineer'd a competitor's hardware/firmware/software product. I am using that knowledge and experience to look at CPAP machines and associated software. I'm having fun!

[cinco777]

I just completed my 30-Day study on the Accuracy of the 7-Day & 30-Day Average AHI, System Leak, and Therapy Usage Hours displayed on the Respironics LCD. Please see my re-edited first posting in this thread for the final results including a new image showing the 30-day completed study/spreadsheet. My conclusion is that the LCD 7-Day and 30-Day Averages are accurate. I wish the Respironics LCD displayed results for 1-day/night but it doesn't. Disappointingly, the newest Respironics unit, PR1, also does not display the results for 1-Day/Night on its LCD. The PR1 does, however, allow a user to set the internal machine clock to local time.

If anyone has questions on the accuracy of the LCD displayed numbers that are not answered by this study, please post to this thread. Thanks.