search

KChen89 / Heartbeat-Classification-based-Diagnosis

Heartbeat classification and labeling
6 stars 3 forks source link

data customization #2

Closed jose4n closed 6 years ago

jose4n commented 6 years ago

Hello again: D, I wonder if it would be possible to customize the type of input data. I am very interested in this project.

I attach the format of data i'd like to process. healthy_person1.txt disease_person1.txt

KChen89 commented 6 years ago

Thanks for your interest! This is a general prototype for research purpose. The data used in this model is sampled at 360Hz and the model is pre-trained with eight types of heatbeat. Size and number of recognizable heartbeat types of the CNN model is fixed since it is a supervised learning model. To customize it for your data. The QRS detection needs more parameter tuning to make it work to a acceptable level and the classification model needs redesign to fit your data sampling rate.

jose4n commented 6 years ago

So,l if i could get data at 360 Hz there will be no problem? I mean, my data ecg values are between 400 and 900, yours from -200 to 200

KChen89 commented 6 years ago

You can try scaling and offset to your signal. Good luck! K.C.