A computer disc drive mfg. company had a problem similar to this irregular heart beat problem, 1985. The disc drive model consisted of 3 modified Lorentz functions. The model was tried on 200 drives; 199 drives agreed with a small standard deviation. But, the last drive showed a -major- defective drive. Thus solving a mfg. problem with a least-square curve fit. So, it can be done ... try it!
See Research Proposal (Human) Heart Beat Math Model found with the CurvFit App ......
and https://www.quora.com/Detection-of-an-Irregular-Human-Heart-Beat-how-to
Hello Phil,
I wouldn't use curve fitting models, they are limited in their ability to detect further aberrations. I would choose wavelet to detect such irregularities.
This is a typical anomaly detection problem.
We've found great success in using Machine Learning tools such as Recurrent Neural Network (RNN) adopting the Long Short Term Memory(LSTM) algorithm. We've demonstrated its sensitivity and diversity on different kinds of heart diseases. We also found great application of the same methodology in other domain disciplines such as predicting and detecting anomaly in jet engines and oil rig machinery offline as well as real time.
Simon Chan nice to hear! Hope others are reading your comments. It can't hurt to try something new. Phil
- Badges
- Science topic
- Similar topics
- Books