Premature Ventricular Complexes (PVCs) and Premature Atrial Complexes (PACs) detector using an ECG-based Deep Learning approach
This repository contains the code for the project of the course Applied AI in Biomedicine at PoliMi.
Read the report for further details.
In this project, we propose a 1D-CNN that is able to predict, in an inter-patient fashion, if a beat is Normal, Premature Ventricular Complex or Premature Atrial Complex relying on only a portion of the ECG. In particular, the CNN requires in input a window length of 500 samples around the R-peak of the beat that we want to predict. This model gives us a quite good result with an accuracy ≥ 0.75 in the worst case related to the PVC class