Decode data files from BMC Medical RESMart GII systems.
Note: the operation of this code has been reverse-engineered from undocumented data. Data outputs may be misunderstood and/or erroneously interpreted.
WARNING: UNFIT FOR ANY MEDICAL USE INCLUDING BUT NOT LIMITED TO DIAGNOSIS AND MONITORING. NO WARRANTY EXPRESS OR IMPLIED. UNDOCUMENTED DATA INTERFACE: USE AT OWN RISK
The RESMart II system records and stores operational data on an SD card. This data is also analyzed and interpreted by the RESmart nPAP Data Analysis Software also made by BMC. Since this is not freely available, I have created this software as an alternative.
Thsi software extracts raw data for your own analysis, including time and duration of use, IPAP, EPAP, and Reslex pressure settings, airflow, tidal volume, respiration rate, as well as SP02 and pulse rate if the auxillary pulse oximeter is used.
Note that the determination of apnea/hypopnea events seems to be done only in the BMC software and not on the device, so this is not currently supported by the software here.
The RESmart GII systems record operational data on an SD card. To extract this data, copy the files using an SD card into your own computer and run the python program.
The files will be in the form:
NNCNNNNN.evt
NNCNNNNN.idx
NNCNNNNN.usr
NNCNNNNN.log
NNCNNNNN.000
NNCNNNNN.001
.
.
.
The NNCNNNN file root is the device serial number, for example '16C01034'
Raw data from the device is stored in the files with numerical extensions. It's my guess that this is analyzed by the BMC software and results are stored in the .usr and .evt files, but without confirmation this is conjecture. User information such as name and address is stored in the .usr file, while the .log file seems to store interactions with the software.
The code here reads only the raw data files. Run in the same directory, it looks for raw data files of the form *.nnn where nnn is a three-digit integer. These are read in sequence.
Each file consists of a sequence of 256-byte packets, Each packet corresponds to one second of data formatted as 120 two-byte unsigned integers followed by an 8-byte timestamp. The format of the integers is as follows (these are guesses!):
Address Interpretatation
000 Always 0xAAAA
001 Reslex value (1-5)
002 IPAP value in units of 0.5 cm H20 (divide by 2 to get cm)
003 EPAP value in units of 0.5 cm H20 (divide by 2 to get cm)
004-028 25 values of something related to pressure at 25 Hz
029-053 25 values of of something related to pressure at 25 Hz
054-078 25 values of instaneous flow at 25 Hz
079-084 10 values of something oscillatory at 10Hz (motor drive?)
085-094 unknown values related to pressure?
095 Tidal volume in liters per minute
098 spO2 (blood oxygenation) in percent (only if oximeter attached)
099 Heart rate in bpm (only if oximeter attached)
100 Respiration rate in breaths per minute
101-120 zero padding
Some values are 0xffff (65535) when not valid, for example the respiration rate takes 30 or more seconds to become valid after the start of pressure flow.
The last 8 bytes are the timestamp, one 16-bit integer for the year, followed by 5 unsigned bytes for month, day, hour, minute, and second. The final byte is an unknown value.
usage: python resmart_parse.py
[in a directory containing .000, .001... raw data files ]
positional arguments:
output_file Output data CSV file
optional arguments:
-h, --help show this help message and exit
--list_dates, -l Print dates found in stored data. Suppresses other
output
--f25_hz, -2 Print out all 25 Hz (flow) data. this will make output
files 25x as big.
--f10_hz, -1 Print out all 10 Hz (pulse) data. this will make output
files 10x as big.
--all_data, -a Print out all 1Hz data fields known or unknown
--time_ymd, -y Print timestamp in Y, M, D, H, M, S format
--time_seconds, -s Print timestamp in seconds since beginning of month
--quiet, -q Do not print progress and info to stderr