In heart failure (HF), detecting the disease at an early stage is crucial, as prompt start of the treatment can significantly increase the quality of life. Continuous monitoring of cardiac conditions using photoplethysmography (PPG) could enable early detection and reduce the burden of heart failure.
Premature ventricular contractions (PVCs) lead to an increase in cardiac contractility, referred to as post-extrasystolic potentiation (PESP). Despite PVCs commonly occurring even in healthy subjects, the recovery of cardiac contractility after a PESP differs based on the severity of HF. In our previous work, we used the maximum amplitude of the PPG signal as a surrogate of cardiac contractility and modeled the recovery in contractility after a PESP using a second-order model. The damping parameter of this model reveals information about the myocardial state and differs depending on the patient's ejection fraction. However, the estimated damping parameter suffers from high variability because the amplitude of the PPG signal is often affected by noise.
In this work, we extend our previous framework exploring whether morphological features derived from pulse decomposition analysis (PDA) can constitute a more stable surrogate of cardiac contractility in our PESP model. In the PDA, each PPG pulse is decomposed into systolic and diastolic phases using two skewed-Gaussian basis functions, and the parameters associated with the systolic phase are used as the morphological features that constitute the input of the PESP model.
The proposed approach, based on morphological features, reduced the coefficient of variation of the damping parameter from 0.34 to 0.28 on average across patients, suggesting a more stable estimate of the myocardial state.
Our model promises to be an effective strategy for assessing cardiac contractility via PPG-based continuous monitoring and, ultimately, enabling the early detection of heart failure.