In a few months I will be conducting a study where we take participents ECGs and try to measure stress levels from that using machine learning. I am also involved in pre-processing the data, meaning anonymizing it, before we enter it into our databases. If possible I would like to run a basic scan over that data before we anonymize it, so if something irregular is detected I can send a recommendation to get that checked out by a professional. Are there any tools readily available for that that do that automatically? If so, which?

  • Almost all ECG machines print an analysis/diagnosis on the output. Are you talking about something more than that?
    – Carey Gregory
    Mar 20, 2019 at 15:14
  • 1
    As a side note, be sure that you have this process documented and approved by your IRB.
    – Bryan Krause
    Mar 20, 2019 at 17:11
  • There are standard values for each of the segments/components of an ECG as far as size, time, etc. These are all on a per lead basis, and as @CareyGregory says, machines already print possible anomalies on an ECG strip when those elements vary from "normal" ranges. For example, on my ECG's the repolarization of the atria gets picked up in some of the leads and appears as a possible bundle branch block (Generally this event is not detectable).
    – JohnP
    Mar 20, 2019 at 17:41
  • If you are asking for the algorithm for detection, this likely belongs in a computing Stack. The rules are complicated and there are enormous databases for some of them, and I don't know of any that aren't proprietary to either the ECG machine company or an academic institution. Hence a computing Stack. I double what Bryan said about IRB, there are a lot of potential issues with collecting a test but not interpreting the results.
    – DoctorWhom
    Mar 21, 2019 at 6:42
  • Well the thing is we will probably use a chest strap or a DIY kit with as little cables/leads as possible for our measurements, since we are probably mostly interested in only HRV and blood-pressure estimation to make it less invasive. So can I expect that a non-professional ECG kit already displays anomalies? I personally only have worked with professional paramedic tools, which we probably won't get for this study, since they are too invasive and to expensive.
    – jaaq
    Mar 26, 2019 at 12:29


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.