A skewed-Gaussian model for pulse decomposition analysis of photoplethysmography signals.

Abstract

Objective: Pulse Decomposition Analysis (PDA) has been proposed to extract reliable information from photoplethysmography (PPG) morphology by decomposing the signal in its physiological sub-waves. The Gaussian model has been widely used in the literature, even though it often underperforms because it is limited to symmetric morphologies. More advanced asymmetric models, such as the Gamma model, have been proposed to achieve improved accuracy. However, the physiological interpretation of the Gamma model is less effective than the Gaussian model, challenging the assessment of the clinical relevance of the outcomes. This paper aims to design an asymmetric PDA model with improved accuracy and effective physiological interpretability.

Approach: We implemented a novel PDA model called the Skewed-Gaussian model and tested it on 8000 PPG pulses from the MIMIC-III Waveform Database. The performances were compared with the reference Gamma-Gaussian model. Models' accuracies were assessed using the residual sum of squares, while Bland-Altman plots were used to evaluate biases. Lastly, the sensitivity and robustness of the models to the initial values' choice were evaluated using random initial values.

Main results: Our model achieved significantly higher accuracy than the reference model. The analysis with random initial values suggested that the model was less sensitive and consistently more robust. Finally, we highlighted the physiological interpretation of the model.

Significance: The proposed model may help to establish a link between alterations in cardiovascular functions and variations detectable in the PPG signal, as well as opening up new avenues for PPG-based remote patient monitoring.