Shape and amplitude decoupling in pulsatile physiological signal synthesis and its evaluation

Area of Science:

• Biomedical Engineering
• Signal Processing
• Cardiovascular Dynamics

Background:

• Pulsatile physiological signals (e.g., arterial blood pressure, ECG) encode cardiovascular dynamics.
• Current generative methods often mix waveform shape and amplitude, limiting control.

Purpose of the Study:

• To introduce VABAM, a framework for controlled physiological signal generation.
• To decouple waveform shape and amplitude for independent manipulation.

Main Methods:

• Developed VABAM, a generative framework using cascaded filtering.
• Introduced novel metrics for evaluating synthesis quality (shape factorization, preservation, controllability, spectral similarity).

Main Results:

• VABAM successfully decouples waveform shape and amplitude.
• The framework outperforms existing methods on benchmark datasets.
• Demonstrated significance of shape-amplitude decoupling in signal generation.

Conclusions:

• VABAM enables controlled physiological signal generation with independent shape and amplitude modulation.
• Potential applications include amplitude-targeted augmentation, uncertainty-quantified prediction, and real-time anomaly monitoring.
• Advances clinical decision-making in physiological signal analysis.