import argparse
import h5py
import numpy as np
import os
from functools import reduce
import pandas as pd
import matplotlib.pyplot as plt
from matplotlib.widgets import Slider, Button, RadioButtons
import sys
from pyqtgraph.Qt import QtGui, QtCore
import torch
from torch.utils.data import TensorDataset,DataLoader
from PyQT_Plot import create_dashboard
from preprocess_data import data_read,windowing_and_resampling_hr,windowing_and_resampling_br
from utils import load_model_HR,load_model_BR,compute_heart_rate
def main(args):
preprocessed_patient_data = data_read(args)
print('-------- Data Acquisition Complete --------')
windowed_patient_overlap,windowed_patient = windowing_and_resampling_hr(preprocessed_patient_data)
print('-------- Pre-processing Complete for HR---------')
windowed_patient_overlap_br = windowing_and_resampling_br(preprocessed_patient_data)
print('-------- Pre-processing Complete for BR---------')
###
patient_ecg = np.asarray(windowed_patient_overlap['ecg'][0][:60])
actual_ecg_windows = np.asarray(windowed_patient['ecg'][0][:60])
patient_ecg_br = np.asarray(windowed_patient_overlap_br['ecg'][0][:60])
###
batch_len = 32
batch_len_br = 1
window_size = 5000
patient_ecg = torch.from_numpy(patient_ecg).view(patient_ecg.shape[0],1,patient_ecg.shape[1]).float()
input_ecg = TensorDataset(patient_ecg)
testloader = DataLoader(input_ecg,batch_len)
patient_ecg_br = torch.from_numpy(patient_ecg_br).view(patient_ecg_br.shape[0],1,patient_ecg_br.shape[1]).float()
input_ecg_br = TensorDataset(patient_ecg_br)
testloader_br = DataLoader(input_ecg_br,batch_len_br)
SAVED_HR_MODEL_PATH = args.saved_hr_model_path
SAVED_BR_MODEL_PATH = args.saved_br_model_path
device = args.device
ecg_peak_locs = load_model_HR(SAVED_HR_MODEL_PATH,testloader,device,batch_len,window_size)
br_peak_locs = load_model_BR(SAVED_BR_MODEL_PATH,testloader_br,device,batch_len,window_size)
### Finding Stored Paths
save_dir = args.save_dir
if not(os.path.isdir(save_dir)):
os.mkdir(save_dir)
save_path = save_dir + '/r_peaks_patient_' + str(args.patient_no) + '.csv'
all_hr = []
initial_hr = len([peak for peak in list(ecg_peak_locs) if peak < 5000 * 6])
for i in range(patient_ecg.shape[0]):
all_hr.append( len([peak for peak in list(ecg_peak_locs) if peak > i * 2500 and peak < (i * 2500 ) + 5000 * 6 ]))
unique = np.unique(np.asarray(all_hr))
peak_no = np.linspace(1,len(ecg_peak_locs),len(ecg_peak_locs)).astype(int)
peak_no = peak_no.reshape(-1,1)
ecg_peak_locs = ecg_peak_locs.reshape(-1,1)
ecg_peak_locs = np.hstack((peak_no,ecg_peak_locs))
pd.DataFrame(ecg_peak_locs).to_csv(save_path , header=None, index=None)
print('-------- R Peaks Saved --------')
all_br = []
initial_br = len([peak for peak in list(br_peak_locs) if peak < 1250 * 6])
for i in range(patient_ecg.shape[0]):
all_br.append( len([peak for peak in list(br_peak_locs) if peak > i * 625 and peak < (i * 625 ) + 1250 * 6 ]))
# all_br.append( len([peak for peak in list(br_peak_locs) if peak > i * 2500 and peak < (i * 2500 ) + 5000 * 6 ]))
i = 1
scatter_peak = []
scatter_peak_1 = []
ecg_point = []
ecg_point_1 = []
k = 0
hr = []
peak_locs = ecg_peak_locs[:,1]
for j in range(len(peak_locs)):
if(peak_locs[j] < 5000*i):
scatter_peak.append(peak_locs[j]-5000*(i-1))
if(i< len(actual_ecg_windows)):
ecg_point.append(actual_ecg_windows[i-1,scatter_peak[k]])
k = k+1
elif(peak_locs[j] >= 5000*i):
scatter_peak_1.append(np.asarray(scatter_peak))
hr.append(compute_heart_rate(scatter_peak_1[i-1]))
ecg_point_1.append(np.asarray(ecg_point))
scatter_peak = []
ecg_point = []
i = i+1
scatter_peak.append(peak_locs[j]-5000*(i-1))
k = 0
if(i< len(actual_ecg_windows)):
ecg_point.append(actual_ecg_windows[i-1,scatter_peak[k]])
k = k+1
import pdb;pdb.set_trace()
if(args.viewer):
create_dashboard(actual_ecg_windows,scatter_peak_1,all_hr,all_br)
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument('--path_dir',help = 'Path to all the records')
parser.add_argument('--saved_hr_model_path',help = 'Path to saved Heart rate model')
parser.add_argument('--saved_br_model_path',help = 'Path to saved breathing rate model')
parser.add_argument('--patient_no',default = 8,type = int,help = 'Patient used for testing')
parser.add_argument('--device',default = 'cuda', help = 'cpu/cuda')
parser.add_argument('--save_dir',default = 'saved_models/',help = 'Directory used for saving')
parser.add_argument('--viewer',default = 0,type = int, help = 'To view ECG plot: 1, else: 0')
args = parser.parse_args()
main(args)
Datasets
Models
