def respiration(data, d_time, frequency=256):
b, a = signal.butter(8,0.05)
# get the raw data and filter it
respiration = signal.filtfilt(b, a, data, padlen=150)
# plt.plot(respiration)
# plt.show()
# we can then get the amplitude and clear signal of the data
# and also the respiration rate
out = resp.resp(respiration,sampling_rate=frequency, show=False)
# plt.plot(out['resp_rate_ts'], out['resp_rate'])
# plt.ylabel('Respiratory frequency [Hz]')
# plt.xlabel('Time [s]');
# plt.show()
# In this case, shape is modify, since we're breaking into frequency
resp_rate_ts = out['resp_rate_ts']
resp_rate = out['resp_rate']
resp_filtered = out['filtered']
# What to Return ? ? ? ?
resp_rate_interpolated = int_spline(resp_rate, resp_rate_ts, d_time)
return lognorm2norm(resp_rate_interpolated)
def main(dataset, debug=False):
ecg = dataset['ECG']
# Detect onset of QRS complexes
qrs = detectQRSPeaks(ecg, False)
# EDR
dt1 = -1
dt2 = 1
edrvals = []
for sample in qrs:
x = getx(ecg, sample+dt1, sample+dt2)
edrvals.append(edr(x))
# Spline interpolation
breathing = interpolate.splrep(qrs, edrvals, s=0)
s = range(0, len(ecg))
interpolation = interpolate.splev(s, breathing, der=0)
[_, filtered, zeros, _, _]= resp.resp(signal=interpolation, sampling_rate=250, show=False)
if debug is True:
# Plot result
resp_ref = dataset['Resp']
import matplotlib.pyplot as plt
plt.plot(qrs, edrvals, 'g', label="raw edr")
plt.plot(s, filtered, 'b', label="spline + filter")
plt.plot(s, resp_ref, 'r', label="reference")
plt.plot(zeros, resp_ref[zeros], 'm*', label="in - out")
plt.legend()
plt.show()
return group_epochs(filtered, zeros, 250, 30)
def getResp_std(task, index, output, samp_rate):
vals = []
for i in range(len(task)):
try:
vals.append(task[i][index])
except:
vals.append(task[i])
vals = filter(lambda x: x != "", vals) # remove empty strings
# extract respiration features using biosppy
# vals is raw resp data
out = resp.resp(signal=vals, sampling_rate=samp_rate, show=False)
# getting instantaneous respiration rate
out_resp = out[4].tolist() # using numpy, convert ndarry into a list
stdev = np.std(out_resp)
output.write(str(stdev) + ', ')
def resp_intervals(self, data):
"""Calculates respiration intervals and indicates inhale/exhale
Parameters
----------
data : breathing signal to be processed
sampling_rate: sampling rate of breathing signal
Returns
----------
interval_lengths: list of breathing interval lenghts
interval_breathe_in: list of breathe in True/False flags
"""
processed_data = resp.resp(signal=data,
sampling_rate=self.srate,
show=False)
filtered_signal = processed_data[1]
inst_resp_rate = processed_data[4]
self.filtered = filtered_signal
self.resp_rate = inst_resp_rate
signal_diff = np.diff(filtered_signal)
signal_signum = signal_diff > 0
resp_changes = np.append(
np.where(signal_signum[:-1] != signal_signum[1:])[0],
[len(signal_signum) - 1])
self.resp_changes = resp_changes
resp_intervals = np.append([0], resp_changes)
interval_lengths = np.diff(resp_intervals)
interval_breathe_in = [signal_signum[i] for i in resp_changes]
self.interval_lengths, self.interval_breathe_in = interval_lengths, interval_breathe_in
last_interval = -1
if len(resp_changes) > 1:
last_interval = resp_changes[-1] - resp_changes[-2]
else:
last_interval = resp_changes[-1]
self.last_breath = last_interval
self.last_is_inhale = signal_signum[resp_changes[-1]]
def resp_intervals(data, sampling_rate=200, last_breath=False):
"""Calculates respiration intervals and indicates inhale/exhale
Parameters
----------
data : breathing signal to be processed
sampling_rate: sampling rate of breathing signal
last_breath : flag that indicates whether single/mult breath analysis is requested
Returns
----------
interval_lengths: list of breathing interval lenghts
interval_breathe_in: list of breathe in True/False flags
"""
processed_data = resp.resp(signal=data,
sampling_rate=sampling_rate,
show=False)
filtered_signal = processed_data[1]
inst_resp_rate = processed_data[4]
signal_diff = np.diff(filtered_signal)
signal_signum = signal_diff > 0
resp_changes = np.append(
np.where(signal_signum[:-1] != signal_signum[1:])[0],
[len(signal_signum) - 1])
if not last_breath:
resp_intervals = np.append([0], resp_changes)
interval_lengths = np.diff(resp_intervals)
interval_breathe_in = [signal_signum[i] for i in resp_changes]
return interval_lengths, interval_breathe_in
else:
if len(resp_changes) > 1:
last_interval = resp_changes[-1] - resp_changes[-2]
else:
last_interval = resp_changes[-1]
return last_interval, signal_signum[resp_changes[-1]]
def getResp_std(task,index,output,samp_rate):
vals = []
for i in range(len(task)):
try:
vals.append(task[i][index])
except:
vals.append(task[i])
vals = filter(lambda x:x !="", vals) # remove empty strings
# extract respiration features using biosppy
# vals is raw resp data
out = resp.resp(signal=vals, sampling_rate=samp_rate, show=False)
# getting instantaneous respiration rate
out_resp = out[4].tolist() # using numpy, convert ndarry into a list
stdev = np.std(out_resp)
output.write(str(stdev) + ', ')
def getRespMax(task, index, output, samp_rate):
vals = []
# making sure we are able to get values whether or not they are in a
# two-dimensional array
for i in range(len(task)):
try:
vals.append(task[i][index])
except:
vals.append(task[i])
vals = filter(lambda x: x != "", vals) # remove empty strings
# extract respiration features using biosppy
# vals is raw resp data
out = resp.resp(signal=vals, sampling_rate=samp_rate, show=False)
# getting instantaneous respiration rate
out_resp = out[4].tolist() # using numpy, convert ndarry into a list
maximum = str(max(out_resp))
output.write(maximum + ', ')
def getRespMax(task,index,output,samp_rate):
vals = []
# making sure we are able to get values whether or not they are in a
# two-dimensional array
for i in range(len(task)):
try:
vals.append(task[i][index])
except:
vals.append(task[i])
vals = filter(lambda x:x !="", vals) # remove empty strings
# extract respiration features using biosppy
# vals is raw resp data
out = resp.resp(signal=vals, sampling_rate=samp_rate, show=False)
# getting instantaneous respiration rate
out_resp = out[4].tolist() # using numpy, convert ndarry into a list
maximum = str(max(out_resp))
output.write(maximum + ', ')
def add_respiration_rate(df: DataFrame) -> None:
""" Process chest movement signal to calculate respiration rate. Add the respiration rate to input as
'respiration_rate' feature.
Args:
df(pandas df): dataset with chest movement signal labelled as 'r'
"""
df["respiration_rate"] = np.nan
all_pilots = df.pilot.unique()
all_experiments = df.experiment.unique()
for pilot in all_pilots:
for experiment in all_experiments:
where_in_df = df.index[(df.pilot == pilot)
& (df.experiment == experiment)]
subset = df.loc[where_in_df, ['time', 'r']]
try:
subset.sort_values(by='time')
out = resp.resp(signal=subset['r'],
sampling_rate=256,
show=False)
where_in_subset = out['resp_rate_ts']
resp_rate = out['resp_rate']
global_ind = where_in_df[where_in_subset]
for i in range(len(resp_rate)):
df.loc[global_ind[i]:global_ind[i + 1],
['respiration_rate']] = resp_rate[i]
except:
print('Not all respiration rates were calculated')
df["respiration_rate"].astype('float32')
# -*- coding: utf-8 -*-
"""
Created on Wed Jul 11 15:32:13 2018
@author: Inki Kim's lab
"""
import os
import pandas as pd
from tqdm import tqdm
from biosppy.signals import ecg,resp
os.chdir('path/to/2_SycronizedData')
outputpath = 'path/to/3_CalcBR/'
for file in tqdm(os.listdir()):
signal = pd.read_csv(file)
out = resp.resp(signal=signal.iloc[:,3], sampling_rate=32, show=True)
ts_br = out['resp_rate_ts']/64
br = out['resp_rate']
df = pd.DataFrame({'time':ts_br})
df['Breathe_rate'] = br*60
df.to_csv(outputpath + file.split('.')[0] + '_BR_2017.csv', index = False)