def plot_raw(path,
emotion_section={
"Stress": [300, 600],
"Amusement": [900, 1200]
}):
arc = OpenSignalsReader(path)
# 心拍データからピークを取り出す
ecg_result = signals.ecg.ecg(signal=arc.signal(['ECG']),
sampling_rate=1000.0,
show=False)
# 皮膚コンダクタンス からSCRを取り出す
eda_filtered = eda_analysis.eda_preprocess(arc.signal(['EDA']), 1000.)
# 呼吸周波数を取り出す
resp_result = resp_analysis.resp(arc.signal('RESP'), show=False)
# 描画設定
fig, axes = plt.subplots(3,
1,
sharex=True,
figsize=(16, 9),
subplot_kw=({
"xticks": np.arange(0, 1200, 100)
}))
axes[0].set_title(path)
# 心拍変動の描画
axes[0].plot(ecg_result['heart_rate_ts'], ecg_result['heart_rate'], 'b')
axes[0].set_xlim(0, 1200)
axes[0].set_ylabel("HR[bpm]")
# 皮膚コンダクタンスの描画
axes[1].plot(arc.t, arc.signal(['EDA']))
axes[1].set_ylabel('[us]')
axes[1].set_ylim(0, 25)
# 呼吸の描画
axes[2].plot(arc.t, arc.signal(['RESP']))
axes[2].set_ylabel('RESP[Hz]')
axes[2].set_ylim(-50, 50)
for i in range(3):
for key in emotion_section.keys():
if key == "Stress":
axes[i].axvspan(
emotion_section[key][0], #start
emotion_section[key][1], #end
alpha=0.3,
color="r",
label="Stress")
elif key == "Amusement":
axes[i].axvspan(
emotion_section[key][0], #start
emotion_section[key][1], #end
alpha=0.3,
color="b",
label="Amusement")
plt.legend()
plt.tight_layout()
plt.xlabel("Time[s]")
return plt
def biosignal_multi_summary(path_list, emotion=None):
i = 0
df_summary = None
for path in path_list:
print('{} ....start'.format(path))
for n in range(3):
arc = OpenSignalsReader(path, multidevice=n)
print("device= {}".format(n))
# 心拍変動
rri_peaks = signals.ecg.ecg(arc.signal('ECG'),
show=False)['rpeaks']
# 呼吸変動
resp_peaks = resp_analysis.resp(arc.signal('RESP'),
show=False)['peaks']
# 皮膚コンダクタンス
scr_data = eda_analysis.scr(arc.signal('EDA'), downsamp=4)
# キーワードを設定
keyword = {'id': i, 'path_name': path}
df = pd.DataFrame([])
df = biosignal_features(rri_peaks, resp_peaks, scr_data, emotion,
keyword)
if i == 0:
df_summary = pd.DataFrame([], columns=df.columns)
# ファイルを結合
df_summary = pd.concat([df_summary, df], ignore_index=True)
i += 1
return df_summary
def biosignal_time_summary(path,
duration=300,
overlap=150,
skip_time=None,
outpath=None):
# 生体データを取得
arc = OpenSignalsReader(path)
# 時間変数を作成
time_ = np.arange(duration, arc.t.max(), overlap)
label_ = time_
if skip_time is not None:
time_ = time_ + skip_time
section_ = zip((time_ - duration), time_)
emotion = dict(zip(label_.tolist(), section_))
# HRV Features
rri_peaks = signals.ecg.ecg(arc.signal('ECG'),
sampling_rate=1000.0,
show=False)['rpeaks']
# RESP Features
resp_peaks = resp_analysis.resp(arc.signal('RESP'),
sampling_rate=1000.0,
show=False)['peaks']
# EDA Features
scr_data = eda_analysis.scr(arc.signal('EDA'),
sampling_rate=1000.0,
downsamp=4)
df = pd.DataFrame([])
# 各生体データを時間区間りで算出
df = biosignal_features(rri_peaks, resp_peaks, scr_data, emotion)
return df
def plot_signal(path):
arc = OpenSignalsReader(path)
# 心拍データからピークを取り出す
ecg_result = signals.ecg.ecg(signal=arc.signal(['ECG']),
sampling_rate=1000.0,
show=False)
# 皮膚コンダクタンス からSCRを取り出す
#eda_result = eda_analysis.scr(arc.signal(['EDA']))
eda_result = eda_analysis.eda_preprocess(arc.signal(['EDA']), 1000.)
# 呼吸周波数を取り出す
resp_result = resp_analysis.resp(arc.signal('RESP'), show=False)
# output
# ECG
np.savetxt(r"C:\Users\akito\Desktop\2019年度発表スライド\ecg_sample.csv",
ecg_result["rpeaks"],
delimiter=",")
# EDA
np.savetxt(r"C:\Users\akito\Desktop\2019年度発表スライド\eda_sample.csv",
eda_result[::25])
# RESP
np.savetxt(r"C:\Users\akito\Desktop\2019年度発表スライド\resp_sample.csv",
np.c_[resp_result['resp_rate_ts'], resp_result['resp_rate']])
# 描画設定
fig, axes = plt.subplots(3,
1,
sharex=True,
figsize=(16, 9),
subplot_kw=({
"xticks": np.arange(0, 1200, 100)
}))
axes[0].set_title(path)
# 心拍変動の描画
axes[0].plot(ecg_result['heart_rate_ts'], ecg_result['heart_rate'], 'b')
axes[0].set_xlim(0, 1200)
axes[0].set_ylabel("HR[bpm]")
# 皮膚コンダクタンスの描画
axes[1].plot(eda_result['ts'], eda_result['filtered'])
axes[1].set_ylabel('SCR[us]')
# 呼吸の描画
axes[2].plot(resp_result['resp_rate_ts'], resp_result['resp_rate'], 'b')
axes[2].set_ylabel('RESP[Hz]')
axes[2].set_ylim(0, 0.5)
for i in range(3):
axes[i].axvspan(300, 600, alpha=0.3, color="r", label="Stress")
axes[i].axvspan(900, 1200, alpha=0.3, color="b", label="Amusement")
plt.legend()
plt.tight_layout()
plt.xlabel("Time[s]")
return plt
def biosignal_summary(path_list, emotion=None, output_path=None):
for i, path in enumerate(path_list):
print(path + ' ....start')
# ファイル名およぶフォルダ名を取得
dict_path = os.path.dirname(path)
fname = os.path.splitext(os.path.basename(path))[0]
# pathから名前と日付に変換する
user = dict_path.split("\\")[-1]
day, time = fname.split("_")[-2:]
date = datetime.datetime.strptime(day + " " + time,
'%Y-%m-%d %H-%M-%S')
# キーワードを設定
keyword = {'id': i, 'path_name': path, 'user': user, 'date': date}
arc = OpenSignalsReader(path)
# 心拍変動
rri_peaks = signals.ecg.ecg(arc.signal('ECG'),
sampling_rate=1000.0,
show=False)['rpeaks']
# 呼吸変動
resp_peaks = resp_analysis.resp(arc.signal('RESP'),
sampling_rate=1000.0,
show=False)['peaks']
# 皮膚コンダクタンス
scr_data = eda_analysis.scr(arc.signal('EDA'),
sampling_rate=1000.0,
downsamp=4)
df = pd.DataFrame([])
df = biosignal_features(rri_peaks, resp_peaks, scr_data, emotion,
keyword)
if i == 0:
if output_path == None:
df_summary = pd.DataFrame([], columns=df.columns)
else:
df_summary = pd.read_excel(output_path)
# ファイルを結合
df_summary = pd.concat([df_summary, df], ignore_index=True)
df_summary.to_excel(output_path)
return df_summary
import eda_analysis
import resp_analysis
import pyhrv.frequency_domain as fd
from opensignalsreader import OpenSignalsReader
import biosignal_plot
path = r"C:\Users\akito\Desktop\test.txt"
arc = OpenSignalsReader(path)
# 心拍データからピークを取り出す
ecg_result = signals.ecg.ecg(signal=arc.signal(['ECG']),
sampling_rate=1000.0,
show=False)
# 呼吸周波数を取り出す
resp_result = resp_analysis.resp(arc.signal('RESP'), show=False)
# 描画設定
fig, axes = plt.subplots(2, 1, sharex=True, figsize=(16, 9))
axes[0].set_title(path)
# Compute NNI series
nni = tools.nn_intervals(ecg_result['rpeaks'].tolist())
filtered = biosignal_plot.detrend(nni, 500)
# 心拍変動の描画
axes[0].plot(ecg_result['heart_rate_ts'].tolist(), filtered, 'b')
axes[0].set_ylabel("HR[bpm]")
# 呼吸の描画
axes[1].plot(resp_result['ts'], resp_result['filtered'])