Can high-frequency ECG fluctuations differentiate between healthy and myocardial infarction cases?

Abstract The Electrocardiogram (ECG) is a standard diagnostic tool for heart's health condition, while the “gold standard” in ECG interpretation is still by observation by the cardiologist. In this article we show that the analysis of the high frequency ECG amplitude fluctuations, which are ignored during the conventional interpretation of the ECG by the cardiologist, regarded as measurement noise, can reveal any deviations from the ideal critical behavior as well as the degree of stability of the heart in stochastic stimulus. By applying the time series analysis method termed “method of critical fluctuations” (MCF) to cardiologist-annotated human ECGs – Physikalisch-Technische Bundesanstalt (PTB) Diagnostic ECG database – we show that it is possible to discriminated between ECGs from healthy and myocardial infarction subjects, as well as to quantify the degree to which they approach the “ideal” critical state or how far from criticality they are, respectively. MCF analysis 100% verifies the characterization of the ECGs belonging to the “Myocardial infarction” class by absence of critical fluctuations. Interestingly, a disagreement with the medical classification of the order of ∼10% was found for the ECGs belonging to the “Healthy controls” set. For the specific ECGs we also find no critical fluctuations, which is a strong deviation from the rest ∼90% of the ECGs belonging to the same class and present critical fluctuations. Moreover, based on a proposed reference ECG, the autocorrelation function can also be used to reveal any differentiation of the studied ECGs from the ideal critical state.