CPAPtalk.com

[quote="cpapdork"]Great, more files to decode...

However, I've got some good news. I've decoded the 002 format enough to reliably retrieve the following event data from the card:

• Periodic Breathing (PB)
• Clear Airway Apnea (CA)
• Obstructed Airway Apnea (OA)
• Hypnopnea (H)
• Respiratory Effort Related Arousal (RE)
• Pressure Pulse (PP)

And thank you sleepy, the PDF you sent from the Encore Viewer was critical to decoding this info. Unfortunately, I need a few more. There are some events in 002 files that I still can't correlate (especially FL, VS, and leak data).

Hi, are you the one writing that program (i've been using for a while ) on the web to view these files? Well, I decoded a bit of the files too, and came up with some other information....

First, your Flow limit, Vibratory Snoring, and Large leak data are all in file 002 the $11(#17) 'normal breathing' event. byte 1 is breathing pressure, byte 2 is Flow Limit, byte 3 is Leak level, and byte 4 is Vibratory Snore.

second, in file 001 also, bytes $24-25(#36-37) are the total time recorded in this 'set'. Stored as a serial #, like the start date/time.

Third, event $12(#18) appears to be 'OFF'
Fourth, I've had a $0b(#11) event appear in my files. No idea what it's for either, but it takes 4 additional bytes of data.

also, it appears the first 16 bytes of all 3 files are the same format.

Do you have any idea what the pressure pulse is for? I have a Remstar pro C-Flex plus (with C-Flex turned off) and I still get all sorts of pressure pulses in my files... I'm ASSUMING the #14($0e) in byte 3 is the pressure it's pulsing? (I know, the pressure change event (#2) is pressure level *10 to cover the 1/2 pressures)

if you have any more information that you've figured out, could you please post it, or send it to me? (other than the stuff in the mediawiki file (as of 3/9/11 anyways )

Thank you BryanC,

I'm following this closely and am glad you are decoding too. Could you help my poor sleepy brain and lay out what you've found in the same format that shows the descriptions on the Onkor site. For instance, he shows Periodic Breathing is in the 002 file and the data looks like this "0fDD00XXXX - Periodic Breathing" where DD is related to duration. You are saying we can find pressure, snore and leak level in 002? That is great... but tonight it makes no sense to me. Sigh.

The Pressure Pulses are so the machine can determine if your apnea is occurring with a clear airway or not. After missing a couple breaths, the machine checks by sending the pulse of pressure.

-- SleepyCPAP

BryanC wrote: Hi, are you the one writing that program (i've been using for a while ) on the web to view these files?

Yes, with some contributions from others. The "development center" for this page is on sourceforge (http://www.sf.net/projects/onkor).

First, your Flow limit, Vibratory Snoring, and Large leak data are all in file 002 the $11(#17) 'normal breathing' event. byte 1 is breathing pressure, byte 2 is Flow Limit, byte 3 is Leak level, and byte 4 is Vibratory Snore. second, in file 001 also, bytes $24-25(#36-37) are the total time recorded in this 'set'. Stored as a serial #, like the start date/time. Third, event $12(#18) appears to be 'OFF'

Wow, great job! I'd love to hear how you decoded this!

if you have any more information that you've figured out, could you please post it, or send it to me? (other than the stuff in the mediawiki file (as of 3/9/11 anyways )

Mediawiki actually has the latest info, as I haven't been able to work on it lately. However, I'm now taking a long break from my real job so I should have more time now. I can start incorporating your new results into Onkor soon.

Would you be ok with putting your input into the onkor mediawiki? You'll need to sign up at sourceforge, tell me your username, and I can add you to the project.

Thanks so much for your input!

Hi BryanC,

Wow, now I'm getting it, at least the most important part (Leak) --
I compared to a night about a month ago when I'd had access to a Windows machine, and had printed from Encore software. The night started with the base rate for no leaks -- and the mask I was using tends to show a line somewhat just over 25 LPM (for the venting rate). Sure enough this is the sequence I saw:
11 78 00 1B 00 (where the 1B = integer 27, which represents 27 LPM). Later in the night I'd spiked just above 50 LPM (lips opened perhaps?) and then the leak line declined rapidly. Here is the highest sequence at about that point in the .002 data file: 11 78 00 3C 00 (where the 3C = integer 60).

My brain is still not getting the rest of it... I don't see my breathing pressure (11cm) in any of the rest of it. The 00 before leak you say is Flow Limit, which would always be zero on my machine since it is the "Pro" not "Auto" and always charts as zero -- that makes sense. And it is true I had no snores marked that night so those are the 00 after the Leak always? I looked on one night I had two VS marked, and with a quick scan of the hex data I found one sequence 11 78 00 1C 01 (where the final 01 shows a VS at that moment) but I may have overlooked the second one.

--SleepyCPAP

cpapdork wrote: Wow, great job! I'd love to hear how you decoded this!

First I went to my DME (That's my equipment people, right?) and had them give me a print out of my charts.
Then I wrote a program (VB2010) to display all the known data from the 002 files. I made it draw vertical lines at all the 'events'. then, just to try to figure out what all the numbers meant, I made each made it draw a line on event $11 normal breathing, byte 3. (honestly, I couldn't for the life of mean figure out what the DD in 11DD00XX00 meant. I kept looking at it...'that looks like the time length.... DD?' but, now I know Duration ) ANYWAYS, trying to figure out what the other 'unknown' events were, I made the program allow me to 'turn off' different events. I also made it print in the debug window all the data for each event. So now, looking at the chart from my program, and the chart from the DME, I came across one where I must have snorted at the end, cause I had 1 Vibratory Snore listed, and the data printing out showed a non-zero number in byte 4! So i thought.. HMMMM... Flow limit seems to be almost always 0, so I'm ASSUMING it's the other byte. (researching it... I SWEAR when I did it last night I saw a .1 flow limitation somewhere, but I can't find it now! Maybe it was in the data the DME 'accidentally' gave me when he switch my SD card?)

cpapdork wrote: Mediawiki actually has the latest info, as I haven't been able to work on it lately. However, I'm now taking a long break from my real job so I should have more time now. I can start incorporating your new results into Onkor soon.

Would you be ok with putting your input into the onkor mediawiki? You'll need to sign up at sourceforge, tell me your username, and I can add you to the project.

Thanks so much for your input!

Absolutely. I typed it all up, and formatted it nicely in an Excel file, then printed it to a PDF, but i don't see any way to post a file on here. Lemme see if I can type it up in a way I can post it on here in text or as an image. (or, I'm bac512 on sourceforge, I could upload it to ya there.... then you just just post a link to it here?)

SleepyCPAP wrote:Hi BryanC,

Wow, now I'm getting it, at least the most important part (Leak) --
I compared to a night about a month ago when I'd had access to a Windows machine, and had printed from Encore software. The night started with the base rate for no leaks -- and the mask I was using tends to show a line somewhat just over 25 LPM (for the venting rate). Sure enough this is the sequence I saw:
11 78 00 1B 00 (where the 1B = integer 27, which represents 27 LPM). Later in the night I'd spiked just above 50 LPM (lips opened perhaps?) and then the leak line declined rapidly. Here is the highest sequence at about that point in the .002 data file: 11 78 00 3C 00 (where the 3C = integer 60).

My brain is still not getting the rest of it... I don't see my breathing pressure (11cm) in any of the rest of it. The 00 before leak you say is Flow Limit, which would always be zero on my machine since it is the "Pro" not "Auto" and always charts as zero -- that makes sense. And it is true I had no snores marked that night so those are the 00 after the Leak always? I looked on one night I had two VS marked, and with a quick scan of the hex data I found one sequence 11 78 00 1C 01 (where the final 01 shows a VS at that moment) but I may have overlooked the second one.
--SleepyCPAP

I looked at my leak limit (maybe it should be called leak level?) for every night in the last month, and it seems very steady until I fall asleep, then it bounces all over the place. (and of course, i notice more apneas happening when it's hitting high points.).. I use a nasal pillow, so I ASSUME I'm rolling my face towards my pillow and pushing it loose or something. I thought I slept with my mouth open, by my last sleep study said I don't.. I guess I could just have loose lips )

As for your breathing pressure, you'll only see that at the very start of the data (or whenever yours changes pressures? its not written all the time) and it'll show up as 02 00 00 6e.
6e is your pressure x 10. so 110 = $6e. (it's kinda hard to put 11.5 in one byte of data...)

does anyone have any of the data files & a screen shot of their data graphed out they could send me to dig a little further into these files?
if you would like to help, please email them to me at bac512@users.sourceforge.net


(Edit-- I listed these headers as RemStar, they're actually Respironics....thanks Jedimark for pointing that out to me..... )
I keep trying to post the stuff I have already here, but the editor keeps taking out all the white space and thus screwing up the columns of everything I just got an idea!
copy and paste this in notepad, or something with a fixed font (like system, I think? where everything is the same size, and evenly spaced.) then replace all the _ with space.
(and yeah, i asked for files again in it.)

If_anyone_out_there_can_send_me_lots_of_data_files_(and_possibly_screen_captures_of_their_data_charted)_i'd_like_to_try_and_figure_more_of_this_out....
_you_can_email_me_here,_or_at_ bac512@users.sourceforge.net
___(yeah,_I_didn't_know_how_to_make_my_email_a_link,_so_I_just_made_it_bold_and_underlined,_you'll_need_to_copy_and_paste_it_to_actually_mail_me_:)_)

my_original_information_came_from
https://sourceforge.net/projects/onkor/
and
https://sourceforge.net/apps/mediawiki/ ... le_Formats

Respironics_Data_Header_Files

____________(001,_002,_and_005)
($01-$02)_______Record_length___
($06)___________File_Type______________1,_2,_5___
($07-0A)________Sequence_#_____________(Filename_minus_extention)
($0B-$0E)_______Start_Time_____________Seconds_=_(#14_*_16777216)_+_(#13_*_65536)_+_(#12_*_256)_+_#11.
__________________________Add_Seconds_to_"01/01/1970_00:00:00"
($0F)___________donno___________________random?


_______________(001_only)
($11)___________?????___________________71/73?
($13)___________Pressure_x_10___________both_this_and_$1E_show_pressure_change?_
__________________________________________Maybe_one_is_min/max_for_flex?
($16)___________Ramp_Time_______________minutes___
($17)___________Ramp_start______________Pressure_x_10
($18)___________bit_operands____________bit_7_($8x)=Cflex_type__________________________$0x=off___$8x=on
________________________________________bit_3_($x8)=Cflex+______________________________$x0=Cflex_$x8=Cflex+
________________________________________bit_0-1($x1-$x2)=Cflex_level____________________$x1-$x3
($19)___________Humidifier_Status_______bit_7_($8L)=on,_($0L)=off.______________________L=level
($1A)___________bit_operands____________bit_7_($8x)=System_One_lock_____________________$0x=off____$8x=on
________________________________________bit_6_($4x)=System_One_Resistance_______________$0x=off____$4x=on
________________________________________bit_3_($x8)=Tube_Size___________________________$x0=22mm___$x8=15mm
________________________________________bit_0-2($x1-$x4)=System_One_Resistance_level_(1-5)___________
($1C)___________bit_operands____________bit_6_($4x)=Auto_On_____________________________$0x=off____$4x=on
________________________________________bit_4_($1x)=Auto_Off____________________________$0x=off____$1x=on
________________________________________bit_3_($x4)=Mask_Alert__________________________$x0=off____$x4=on
________________________________________bit_2_($x2)=Show_AH/I___________________________$x0=off____$x2=on
($1E)___________Pressure_x_10___________both_this_and_$13_show_pressure_change?_
__________________________________________Maybe_one_is_min/max_for_flex?
($24-$25)_______Record_length___________Seconds._(#25_*_256)_+_#24.
($2C)___________donno___________________always_0_or_1
($31)___________donno___________________always_0_or_22
($38)___________donno___________________always_0_or_2
($42-$45)_______donno___________________Seems_random,_but_2_files_done_within_minutes_have_similar_numbers
($48-$4C)_______donno___________________Seems_random?
_______________(005_only)
($10)___________donno___________________always_0_or_1
($17)___________donno___________________random?


-_______________didn't_find_____________Humidifier_LED__________off/on
-_______________didn't_find_____________Language________________EN/ES
-_______________didn't_find_____________Tube_Type_Lock__________off/on
-_______________didn't_find_____________Ramp_button_backlight___off/on

Respironics_Record_Events

(Header record only ONCE at the beginning of the 002 file)
002_____(hex)____________________Data____Byte_1_________Byte_2________Byte_3________Byte_4__________Byte_5
__________02_Set_Pressure__________3_____Duration(0)____Flow_limit____0____________Pressure_*_10___
__________04_Pressure_Pulse________3_____Duration_______Flow_limit____0___________
__________05_RERA__________________3_____Duration_______Flow_limit____Leak_Limit
__________06_Central_Apnea_________3_____Duration_______Flow_limit____Leak?Limit
__________07_Obstructive_Apnea_____3_____Duration_______Flow_limit____Leak?Limit
__________0A_Hypnopnea_____________3_____Duration_______Flow_limit____Leak?Limit
__________0B_unknown_______________4_____Duration_______Flow_limit____????__________???
__________0C_unknown_______________3_____Duration_______Flow_limit____????
__________0D_unknown_______________2_____Duration_______Flow_limit
__________0E_unknown_______________5_____Duration_______Flow_limit____????__________0-5?_____________0/60?
__________0F_Periodic_breathing____5_____Duration_______Flow_limit____????__________???______________??
__________10_unknown_______________5_____Duration_______Flow_limit____????__________0?_______________90?
__________11_Normal_breathing______4_____Duration_______Flow_limit____Leak_Limit____Vsnore___
__________12_End___________________5_____Duration(0)____Flow_limit____2/1?__________VSnore___________??

(Header_record_comes_before_every_1502_byte_block)
005_____0-$5DC_(1500)__Breathing_pressure__(This_can_be_0_to_1500_bytes.
_________________________Take_header_bytes_1_&_2,_subtract_26,_and_that's_the_record_Length.___
_________________________Header_repeats_with_each_record)
________$5DD-$5DE______16_bit_checksum?__Actual_position_changes_with_record_size._(i.e._your_last_record_may_only_be_30_bytes,_then_it_would_be_bytes_31_and_32.)

Posting this here so it's documented somewhere for anyone else who's interested..

I haven't looked at the 001 file yet, as I haven't needed to yet. It's mostly been done by others anyway.

This is the result of my hackings built on top of the hard work of others.

.002 File Format Header

02 [xx xx xx xx] 02 [02] [xx xx xx xx] [xx xx xx xx] [xx]

[1] = 32 bit LE File Length, (it's possible this is 16bit and the second set of bytes is version -> Test to prove it.. Run the machine for several days straight)
[6] = File Type (same as extension)
[7] = 32bit LE Sequence (same as filename)
[11] = 32bit LE UNIX Timestamp, seconds since 1/1/1970 0.0.0
[15] = 8bit Sum of bytes 0..14

the 02's are respironics magic number.. one of them possibly represent file format version.

Header is followed by raw event data of filelength-16-2, and a 2 byte checksum at the end.

.005 Waveform Header, appears before each section

02 [XX XX] [XX XX] 02 [05] [XX XX XX XX] [XX XX XX XX] [XX XX] [XX] [XX XX] [XX XX] [??] [XX]

[1] = 16bit LE block length
[3] = 16bit LE field, likely a version field. I have seen values other than 1
[6] = File Type (same as extension)
[7] = 32bit LE Sequence (same as filename)
[11] = 32bit LE UNIX Timestamp
[15] = 16bit LE Number of Seconds in this block
[17] = 8bit ==1 ?? Unneeded
[18] = 16bit LE ==1 ?? Unneeded
[20] = 16bit LE ==5 Samples per second
[22] = 8bit ==0 ?? Unneeded
[23] = 8bit sum of previous header bytes

LE==Little Endian
Each .005 chunk header is followed by blocklength-2-24 bytes of sample data, followed by a 2 byte checksum. Cycle repeats until end of file.
Sample data is stored 8bit Signed.

.002 Event Structure:
Event records are variable lengths, and start with a code between 0x00 and 0x12

0x00 ??Unknown [16bit LE DELTA],[8bit UNKNOWN]
0x01 ??Unknown [16bit LE DELTA],[8bit UNKNOWN]
0x02 PRESSURE [16bit LE DELTA],[8bit PRESSURE] *Divide pressure by 10.0
0x03 BIPRESSURE [16bit LE DELTA],[8bit PRESSURE_LOW],[8bit PRESSURE_HIGH] *Divide pressures by 10.0
0x04 PRESSUREPULSE [16bit LE DELTA],[8bit ???] *Have observed values 14 and 15 for 8bit data field.
0x05 RERA [16bit LE DELTA],[8bit OFFSET]
0x06 OBSTRUCTIVE [16bit LE DELTA],[8bit OFFSET]
0x07 CLEARAIRWAY [16bit LE DELTA],[8bit OFFSET]
0x08 ??? - Not Observed Yet <-- assume [16bit LE DELTA], [8bit Unknown] ?
0x09 ??? - Not Observed Yet <-- assume [16bit LE DELTA], [8bit Unknown] ?
0x0A HYPOPNEA [16bit LE DELTA],[8bit OFFSET]
0x0B ??? - Unknown [16bit LE DELTA], [8bit Unknown], [8bit Unknown]
0x0C FLOWLIMITED [16bit LE DELTA],[8bit OFFSET]
0x0D VIBRATORYSNORE [16bt LE DELTA]
0x0E ??? - Unknown [16bit LE DELTA], [8bit Unknown], [8bit Unknown], [8bit Unknown]
0x0F CSR [16bit LE DELTA], [16bitLE DURATION],[8bit OFFSET]
0x10 ??? - Unknown [16bit LE DELTA], [16bit LE Unknown], [8bit Unknown]
0x11 LEAKRATE [16bit LE DELTA], [8bit LEAKVALUE], 8bit[VIBRATORYSNORE] *Secondary type of Vibratory snore when >0
0x12 SUMMARY [8bit Unknown], [8bit PRESSURE], [16bit Unknown] * Pressure summary information. Only occurs at the end of event file.

Unknown fields are not needed to replicate Encore functionality.
The secondary Vibratory Snore thing is why Encore reports more Vibratory Snore's in the sleep flags than overlayed on the waveform. Not sure what they really mean by it.
I haven't interpreted LEAKRATE 100%. Basically I subtract 18 from it and get graphs that visually match encore. Not scientific enough for me to be comfortable with that explanation though.

Time calculations:
DELTA and OFFSET fields are in seconds.
When parsing this event data, take the initial UNIX timestamp from the header as the starting point
Add each DELTA sequentially to the time _before_ using the event record.
Subtract the OFFSET from a copy of the running sequential time to find the start time of the individual event. ie. do not subtract the OFFSET adjustment from the actual running DELTA.. Record this temp value for the individual event records themselves.
It makes sense to only use UTC time functions, and convert to localtime when required for display purposes.

BryanC wrote:does anyone have any of the data files & a screen shot of their data graphed out they could send me to dig a little further into these files?

Bryan,

Have you looked at the known datasets for verification on the onkor wiki? there are currently two full 001/002/005 datasets with the complete Encore graphs for them. I have several others that h4lpful participants on this forum have sent me that I can post after "sanitizing" of personal information.

See http://sourceforge.net/apps/mediawiki/o ... rification

001 file appears appears to contain initial settings info. See wiki again or the actual code:
http://onkor.hg.sourceforge.net/hgweb/o ... on.py#l153

cpapdork wrote:
Bryan,

Have you looked at the known datasets for verification on the onkor wiki? there are currently two full 001/002/005 datasets with the complete Encore graphs for them. I have several others that h4lpful participants on this forum have sent me that I can post after "sanitizing" of personal information.

See http://sourceforge.net/apps/mediawiki/o ... rification

001 file appears appears to contain initial settings info. See wiki again or the actual code:
http://onkor.hg.sourceforge.net/hgweb/o ... on.py#l153

Yeah cpapdork, I got the datasets off your site... And that's where i got all my original information from... I copied your info wrong and had obstructive apnea's listed as event #7 and central Apnea's listed as #6, which is backwards ...

NOW though I'm getting upset looking over all MY sleep records, it seems event #14 happens every night a few minutes before i turn my machine off, so I'm assuming that's what's waking me up.... but, of course, #14 is 'Unknown'

ohh and for those of you following along, looks like some of my layout was 'less correct' than Jedimark's. Though, I'm still unsure about his '16bit LE DELTA' and only having an '8bit Offset'... I think the second byte of his '16bit LE DELTA' is reserved for future use, or something else. (I'm sure I've seen it a 1 sometimes....but I can't seem to find that anywhere anymore.).... and of course, as i tell him all the time, I've only been looking at a 'good set' of data for about 2 weeks. (I've only had this new machine 4 weeks....)

Ok, here is another question for any one:

How can you view how many times you quit breathing (or have apeana's) with this Philips Respironics System One without the software? My doctor has turned the pressure down due to me swallowing the air, and I would like to know how many times I quit breathing at night; this is the main concern I have right now.

RobDunham wrote:Ok, here is another question for any one:

How can you view how many times you quit breathing (or have apeana's) with this Philips Respironics System One without the software? My doctor has turned the pressure down due to me swallowing the air, and I would like to know how many times I quit breathing at night; this is the main concern I have right now.

goto http://www.hoolehan.com/onkor/ cpapdork has a project running that you can upload your data files to and look at what events happened throughout the night.

Hi,

I tried uploading and looking at the output about a month ago but, unfortunately at the time, it did not provide a meaningful output for me. you see, I am using a "PR System One BiPAP Auto" and the routines were set up for an APAP.

My question is, has the web site been updated to deal with BiPAP Auto or is it still only working for CPAP and CPAP Auto?


Cheers

builta wrote:Hi,

I tried uploading and looking at the output about a month ago but, unfortunately at the time, it did not provide a meaningful output for me. you see, I am using a "PR System One BiPAP Auto" and the routines were set up for an APAP.

My question is, has the web site been updated to deal with BiPAP Auto or is it still only working for CPAP and CPAP Auto?


Cheers

ohh, that i couldn't tell ya, you'll have to wait for cpapdork to reply.... BUT, is there any way you could email me a bunch of your files? I have the PRS1 CPAP Flex+ Pro (I believe? only had it about a month, but I know it's not a BiPap)... I'm sure yours has a lot of data I haven't got 'details' for yet, and it would help me figure out how to support the BiPap machines too...
If you can, I need the 001, 002, and 005 files (and, if there are others? (003, 004, 007, etc.) for each? (at this point, i don't know!))
bac512@users.sourceforge.net

Thanks BryanC, that was very helpful!

When I look at the "Events" tab at the "OA", is that how many times I quit breathing?

RobDunham wrote:Thanks BryanC, that was very helpful!

When I look at the "Events" tab at the "OA", is that how many times I quit breathing?

well, partially. OA's and CA's are when you quit breathing. OA's are when something in your throat(?) blocks your airway, and CA's are when you just 'forget' to breath (and unless excessive, I believe even non-cpap people occasionally 'forget' to breath every now and then...).. Hypnopnea means you're breathing, but not very deeply. and the RERA's means it thinks something in your breathing woke you....

from further research I've done on this, if you look at a small enough section of the graph (like click on it, and just select maybe a 1/8 to a 1/4 inch on a 8 hour sleep file) and you'll see the wave form of your breathing. if it looks like a nice steady up and down, kinda like little hills, you're awake. If it looks like it has a little ledge about halfway up, you're asleep. (obviously, there's a LOT of different breathing patterns, so that's not gunna show ya 100%, and it may not be 100% accurate, but after i read it, I looked over mine, and it seemed to be at least accurate for the sections i checked... not sure how fallin in and coming out of sleep look..... but....)