PPG_Clustering

table of most important contents

LDA and Autoencoder.ipynb
this notebook clusters peak to peak ppg signals using LDA and fully connected autoencoder as features extractor. datasets from Dr mostafa peak2peak_aug_2022
general_pca_kmeans_notebook.ipynb
this notebook cluster complete signals of ppg using PCA [2 PCA components] as a features reduction technique [no plots of signals and results was not satisfied]
heart_rate_equqlization.py
this notebook contains functions able to divide PPG signal to peak to peak parts and then normalize each on beat to equal number of time steps
peak2peak_aug_2022_oneBeat_kmeans.ipynb
this notebook contain clustering of Peak_to_peak data from Dr mostafa. algorithms used are DTW from tslearn library. for not good clusters there is subclustering and plotting
peak2peak_aug_2022_oneBeat_kmeans_1.ipynb
this notebook uses PCA but contains error of shuffling
peak2peak_aug_2022_oneBeat_softDTW.ipynb
this notebook uses softDTW to cluster first 50000 of peak2peak datasets but no plots and no blood pressure available because of the memory leakage
revese_approach.ipynb this notebook proof that ppg signal shape not depend on blood pressure only but there are another factors such as heart rate
==>do not forget the advanced-ppg-clustering-peak-to-peak-aug-data respository Link