chaffim/ardupi-ecg

Full HRV analysis of Arduino pulse sensor, using Python signal processing and time series techniques. Chaotic, Fourier, Wavelet, Regression, Neural Net.

ArduPi-ECG

• toolkit for full HRV (heart rate variability) analysis of Pulse Sensor data using standard (Linear Regression, Fourier Extrapolation) compared to groundbreaking new techniques (Wavelet transforms, Chaotic Analysis, Neural Networks)
• script for real-time python display of bpm data from Arduino Pulse Sensor, on Arduino Uno, using Matplotlib.

Requirements: Arduino Uno/Arduino IDE or Raspberry Pi, Python 3, Matplotlib, Numpy, PySerial, Arduino Pulse Sensor

TODO:

• Use NN regression on regressive output from real data (either standard or NN LR) ...?
• Use ALL models on real data
• Build Chaotic model
• Build DeepNNs

Relevant papers:

• Fourier Extrapolation

  • https://www.ncbi.nlm.nih.gov/pubmed/10847190
  • https://www.ncbi.nlm.nih.gov/pubmed/9842406
  • https://www.ncbi.nlm.nih.gov/pubmed/16387047

• Wavelet Extrapolation

  • https://hal.archives-ouvertes.fr/hal-00414210/document
  • http://epubs.siam.org/doi/abs/10.1137/0916005
  • http://geodus1.ta.tudelft.nl/PrivatePages/C.P.A.Wapenaar/6_Proceedings/Soc.Expl.Geoph/Seg_94b.pdf

• Chaotic Analysis:

  • For HRV:
    • http://www.scielo.br/scielo.php?script=sci_arttext&pid=S2358-04292016000500005
    • https://www.ncbi.nlm.nih.gov/pubmed/17593181
    • http://geoffboeing.com/2015/03/chaos-theory-logistic-map/
    • https://www.researchgate.net/publication/306226253_Visual_Analysis_of_Nonlinear_Dynamical_Systems_Chaos_Fractals_Self-Similarity_and_the_Limits_of_Prediction
  • For EKG:
    • https://math.dartmouth.edu/archive/m53f07/public_html/proj/MehtaMiller.pdf
    • http://aip.scitation.org/doi/10.1063/1.166330

• Deep NN:

  • Deep NN with Wavelet
    • HRV and BPV neural network model with wavelet based algorithm calibration: https://pdfs.semanticscholar.org/8e80/4c4fb5efdce51bbdfa5c26930e8a181ddd62.pdf
  • Deep NN with Fitzhugh-Nagumo
    • Deep neural heart rate variability analysis: https://arxiv.org/pdf/1612.09205.pdf
  • Deep NN with Fourier:
    • Training Deep Fourier Neural Networks To Fit Time-Series Data: https://arxiv.org/pdf/1405.2262.pdf

Visualization Tools:

• Linear, Fourier, Wavelet
  • Matplotlib
• Chaos
  • Pynamical - nonlinear visualization - https://github.com/gboeing/pynamical
• NN LR, Deep NN
  • Tensorboard - https://github.com/tensorflow/tensorflow/blob/r1.1/tensorflow/tensorboard/README.md