#! /bin/sh
# file: Make-Data G. Moody October 2008
# This shell script creates the files containing RR interval time
# series from 15 PhysioBank records. See the file "RECORDS", in this
# directory, for information about the original PhysioBank records
# from which these time series come.
# The time series were chosen for use by participants in the Chaos
# controversy issue, "Is the Normal Heart Rate Chaotic?". See
# http://physionet.org/challenge/chaos/
# for details.
#
# Except for standard utilities available on any Unix(-like) platform, the
# software used by this script comes from the WFDB and plt software packages,
# available at
# http://physionet.org/physiotools/wfdb.shtml
# and
# http://physionet.org/physiotools/plt/
# First, make the "unfiltered" time series. See
# http://physionet.org/physiotools/wag/ann2rr-1.htm for a description of
# ann2rr's options as used below.
ann2rr >n1rr.txt -r nsrdb/16265 -a atr -c -i s3 -v s3 -w -f 0 -t 22:14:0
ann2rr >n2rr.txt -r nsrdb/16272 -a atr -c -i s3 -v s3 -w -f 0 -t 23:26:0
ann2rr >n3rr.txt -r nsrdb/16786 -a atr -c -i s3 -v s3 -w -f 0 -t 23:20:0
ann2rr >n4rr.txt -r nsrdb/16795 -a atr -c -i s3 -v s3 -w -f 0 -t 20:45:0
ann2rr >n5rr.txt -r nsrdb/19090 -a atr -c -i s3 -v s3 -w -f 0:35 -t 19:13:0
ann2rr >c1rr.txt -r chfdb/chf01 -a ecg -c -i s3 -v s3 -w
ann2rr >c2rr.txt -r chfdb/chf03 -a ecg -c -i s3 -v s3 -w
ann2rr >c3rr.txt -r chfdb/chf07 -a ecg -c -i s3 -v s3 -w
ann2rr >c4rr.txt -r chfdb/chf08 -a ecg -c -i s3 -v s3 -w
ann2rr >c5rr.txt -r chfdb/chf12 -a ecg -c -i s3 -v s3 -w
ann2rr >a1rr.txt -r ltafdb/11 -a qrs -c -i s3 -v s3 -w -f 11:0 -t 20:10:0
ann2rr >a2rr.txt -r ltafdb/12 -a qrs -c -i s3 -v s3 -w -f 1:0 -t e
ann2rr >a3rr.txt -r ltafdb/15 -a qrs -c -i s3 -v s3 -w -f 5:10 -t 23:57:0
ann2rr >a4rr.txt -r ltafdb/17 -a qrs -c -i s3 -v s3 -w -f 1:0 -t 24:31:0
ann2rr >a5rr.txt -r ltafdb/18 -a qrs -c -i s3 -v s3 -w -f 1:0 -t 24:28:0
# Next, make the "filtered" time series. In sinus rhythm, "nguess"
# replaces abnormal beats and fills gaps with "Q" annotations, and
# ann2rr's -p and -P options cause it to remove all intervals not
# bounded by "N" annotations.
nguess -r nsrdb/16265 -a atr
ann2rr >n1nn.txt -r nsrdb/16265 -a nguess -c -i s3 -v s3 -w -p N -P N \
-f 0 -t 22:14:0
nguess -r nsrdb/16272 -a atr
ann2rr >n2nn.txt -r nsrdb/16272 -a nguess -c -i s3 -v s3 -w -p N -P N \
-f 0 -t 23:26:0
nguess -r nsrdb/16786 -a atr
ann2rr >n3nn.txt -r nsrdb/16786 -a nguess -c -i s3 -v s3 -w -p N -P N \
-f 0 -t 23:20:0
nguess -r nsrdb/16795 -a atr
ann2rr >n4nn.txt -r nsrdb/16795 -a nguess -c -i s3 -v s3 -w -p N -P N \
-f 0 -t 20:45:0
nguess -r nsrdb/19090 -a atr
ann2rr >n5nn.txt -r nsrdb/19090 -a nguess -c -i s3 -v s3 -w -p N -P N \
-f 0:35 -t 19:13:0
# See http://physionet.org/physiotools/wag/nguess-1.htm for an
# explanation of the -m option used below. The effect is to reject
# outliers in these CHF time series, which include many sinus pauses,
# more aggressively than the default strategy used for the healthy
# cases above.
nguess -r chfdb/chf01 -a ecg -m 1.6
ann2rr >c1nn.txt -r chfdb/chf01 -a nguess -c -i s3 -v s3 -w -p N -P N
nguess -r chfdb/chf03 -a ecg -m 1.6
ann2rr >c2nn.txt -r chfdb/chf03 -a nguess -c -i s3 -v s3 -w -p N -P N
nguess -r chfdb/chf07 -a ecg -m 1.6
ann2rr >c3nn.txt -r chfdb/chf07 -a nguess -c -i s3 -v s3 -w -p N -P N
nguess -r chfdb/chf08 -a ecg -m 1.6
ann2rr >c4nn.txt -r chfdb/chf08 -a nguess -c -i s3 -v s3 -w -p N -P N
nguess -r chfdb/chf12 -a ecg -m 1.6
ann2rr >c5nn.txt -r chfdb/chf12 -a nguess -c -i s3 -v s3 -w -p N -P N
# The "nguess" application cannot be usefully applied to atrial fibrillation,
# so it's not used for these 5 time series.
ann2rr >a1nn.txt -r ltafdb/11 -a qrs -c -i s3 -v s3 -w -p N -P N \
-f 11:0 -t 20:10:0
ann2rr >a2nn.txt -r ltafdb/12 -a qrs -c -i s3 -v s3 -w -p N -P N \
-f 1:0 -t e
ann2rr >a3nn.txt -r ltafdb/15 -a qrs -c -i s3 -v s3 -w -p N -P N \
-f 5:10 -t 23:57:0
ann2rr >a4nn.txt -r ltafdb/17 -a qrs -c -i s3 -v s3 -w -p N -P N \
-f 1:0 -t 24:31:0
ann2rr >a5nn.txt -r ltafdb/18 -a qrs -c -i s3 -v s3 -w -p N -P N \
-f 1:0 -t 24:28:0
# Generate a set of plots for visual review of the time series.
for i in *.txt
do
R=`basename $i .txt`
plt $i 2 0 -X 0 100000 -Y 0 2 -x "Elapsed time (seconds)" \
-y "RR interval (seconds)" -t $i -T lw | lwcat -pdf >$R.pdf
done
# Generate MD5, SHA1, and SHA256 checksums that can be used to verify that
# these files have not been altered since they were generated.
md5sum *.txt *.pdf >MD5SUMS
sha1sum *.txt *.pdf >SHA1SUMS
sha256sum *.txt *.pdf >SHA256SUMS