ECG_scans_machine_learning

We've always wanted to get into ml and this provided a perfect platform to utilise it for the greater good

It can predict arrhythmias in the heart from the ecg scan of a patient (data has to be in the right form)

A lot of time was spent preprocessing data into a NN-friednly form. We then used the Tensorflow library to construct NN to make predictions

The size of the data was our biggest challenge. We also couldn't get our recurrent neural network to have a loss not equal to nan.

We created a successful CNN model with a success rate of ~70%

How to manage such large data-sets and download/upload stuff efficiently

We want to successfully construct our RNN, which is better designed for time-series data. We also would like to input meta-data into our NN, like age or gender of the patient

Built With

• numpy
• os
• python
• tensorflow