Mutual corrected conditional entropy can measure RT-RR uncoupling in myocardial infarction patients

The dependence of RT beat-to-beat variability on heart period may involve complex relationships based on an unknown number of previous RR intervals. A mutual corrected conditional entropy, MCCE(L), may provide a global index of dependence, that avoids any a priori model and is a function of the number of heart beats considered. MCCE(L) is implemented through counts of sequences formed by one RT and L-l preceding RR intervals and contains a maximum entropy correction based on the percent of sequences appearing only a single time. The minimum of MCCE(L) indicates the optimal length of sequences in relation to the available data. The difference between its normalised value and l is an index of the regularity induced by RR on RT, ρ_rt-rr. Series of 300 RR and RT were extracted from ECG in 10 normal subjects (NS) and 10 patients two weeks after myocardial infarction (MI). A high ρ_rt-rr was displayed by NS, while this feature appeared to be disrupted in MI