def beat_loc_and_signal(address, begin, end):
file_beat = json.load(open(address + '/beat_loc.json'))[begin:end]
# raw = CsvLoader.load_dummy(1)
file_raw = CsvLoader.load(address + '/record.csv')[begin:end]
return file_beat, file_raw
import CsvLoader import matplotlib.pyplot as plt import numpy as np from scipy.signal import lfilter, convolve data_source = CsvLoader.load()[:600] ''' LOW PASS FILTERING ''' b=[1,0,0,0,0,0,-2,0,0,0,0,0,1] a=[1,-2,1] lowed = lfilter(b,a, data_source) ''' HIGH PASS FILTERING ''' b = [-1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,32,-32,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1] a = [1,-1] highed = lfilter(b,a, lowed) ''' DERRIVIATE ''' h = [x/8. for x in [-1, -2, 0, 2, 1]] derr = convolve(highed, h)[2:] # cancel delay abss = [abs(x) for x in derr] max_abs = max(abss) derr = [x/max_abs for x in derr] '''
import CsvLoader
source = CsvLoader.load()
"""
Delay data: 12 second (2400 sample)
Sample Rate: 200/second (5ms)
Using PanTom
1. import PanTom.py
2. instantiate Detector object
3. call function add_data(raw)
return:
check the returned[0] if
['filling']:
function return 'filling', detection step is skipped, algorithm still filling buffer
[Filtered Data, PVC, PAC, BUNDLE_BRANCH, AtrialTachycardia, VentricularTachycardia, BundleBranchBlock] :
[0] is filtered data, send to chart via MQTT
[1..6] is classification
"""
from PanTom import *
detector = Detector()
for raw in source:
# detector.add_data(raw, True) # view plot
detector.add_data(raw) # deployment
# dat = detector.add_data(raw) # print test, return
# if dat[0] != 'filling':
# print dat[0], dat[1]
# Test sample
# numbers = [100] # normal
numbers = [106] # many PVC
INDEX_LIMIT = 1000
def build_message(index, sample):
next_index = 1 if index + 1 > INDEX_LIMIT else index + 1
return next_index, str(index) + ':' + str(sample)
if __name__ == '__main__':
client = mqtt.Client(client_id=CLIENT_ID)
client.connect("localhost", 1883, 60)
for number in numbers:
number = 'TA-Data/MIT_BIH/' + str(number)
index = 0
print(number)
# raw = CsvLoader.load(number + '/record.csv')[:2935] # 8 second
# raw = CsvLoader.load(number + '/record.csv')[:12000] # 1 minute
raw = CsvLoader.load(number + '/record.csv')[9360:] # cut up
# raw = CsvLoader.load(number + '/record.csv') # till end
for data in raw:
index, message = build_message(index, data)
client.publish(CLIENT_ID + "/sensor", message)
sleep(0.005)
# # for ecg in ecg_h:
# # print ecg
#
# # # ax(3) = subplot(323)
# # fig, (ax_orig, ax_filter) = plt.subplots(2, 1, sharex=True)
# # ax_orig.plot(data_source)
# # ax_orig.set_title('Original pulse')
# # ax_filter.plot(ecg_h)
# # ax_filter.set_title('Band Pass Filtered')
# # fig.tight_layout()
# # plt.show()
import matplotlib.pyplot as plt
import CsvLoader
from PanTom import *
detector = Detector()
source = CsvLoader.load()
for data in source:
dat = detector.add_data(data, True)
if dat[0]!='filling':
print dat[0], dat[1]
# fig, (ax_orig, ax_filter) = plt.subplots(2, 1, sharex=True)
# ax_orig.plot(data_source[:200])
# ax_filter.plot(res)
# plt.tight_layout()
# plt.show()
import paho.mqtt.client as mqtt
import CsvLoader
import time
from PanTom import Detector
detector = Detector()
CLIENT_ID = '02WXO01'
data_csv = CsvLoader.load()
client = mqtt.Client(client_id=CLIENT_ID)
client.connect("localhost", 1883, 60)
# device_id = msg.topic.split('/')[1]
device_id = "02WXO01"
for raw in data_csv:
data = detector.add_data(float(raw))
if data[0] != 'filling':
# send bpm for the first time
bpm = round(float(sum(data[7])) / len(data[7]), 2)
# client.publish("bpm/"+device_id, bpm, qos=2) # ini bikin errorr
client.publish("bpm/" + device_id, bpm)
for filtered_data in data[0]:
print "forwarding data to dashboard", filtered_data
# client.publish("visual/"+device_id, filtered_data, qos=2)
client.publish("visual/" + device_id, filtered_data)
time.sleep(0.010)