Motivation and aim: The adaptation time of the QT interval to sudden abrupt heart rate (HR) changes modeled as a first-order system response an ECG marker to stratify patients for arrhythmic risk. In such a first-order model, this time constant, tau, is the same as the delay in responding to a ramp-like HR change. Thus, the time lag between the actual QT series and the expected memoryless HR-dependent QT series can be estimated from the exercise and recovery phases, being much more feasible than from sudden step-like HR changes.
Materials: 39 stress test ECG recordings were selected from FINCAVAS study and divided into three groups, according to their likelihood for Coronary Artery Disease (CAD): low-risk, mild-risk and high-risk patients.
Methods: The memoryless expected HR-dependent QT interval time series was calculated by fitting four regression models to the [QT(n);RR(n)] data pairs in three stationary QT-to-RR dependent windows. The time lag, tau_*, was estimated by a Mean Square Error fit between the real QT(n) response ramp and the memoryless expected HR-dependent QT(n-tau_*) ramp, separately in exercise, tau_e , and recovery, tau_r , phases.
Results and conclusion: The hyperbolic model provided the best fitting, and therefore, the selected one. The average time lags for exercise and recovery phases (tau_e = 26.65s and tau_r = 69.97s, respectively, in the low-risk group) are in line with those of a previous study estimated from step-like HR changes in daily activities (tau_e = 34.79s and tau_r= 48.40s). Results show that the delay tau_e increases for higher risk patients; and the difference, Delta_tau, between delays tau_e and tau_r is remarkably larger for low-risk patients, being much reduced when CAD risk increases. Interestingly, the delay tau_r manifests a significantly reverted behavior. In conclusion, the proposed markers tau_e, tau_r and Delta_tau show potential for CAD risk stratification.