Non-Contact Blood Pressure Estimation Using Optimized PTT Signal Processing and Chaotic Dynamics

Accurate and continuous blood pressure (BP) monitoring is vital for ensuring cardiovascular health; however, traditional cuff-based methods are impractical for long-term monitoring scenarios. Pulse Transit Time (PTT)-based approach provides a promising alternative for non-contact BP estimation, but it suffers from motion artifacts and environmental noise. This paper presents an optimized PTT extraction method combined with chaotic dynamics modeling to improve BP prediction. Our approach includes multi-stage signal filtering, phase-difference-based PTT computation, and state-space reconstruction for long-term BP estimation. Additionally, ensemble chaos modeling enhances prediction stability. We evaluate our method on the BP-Video dataset, achieving a mean absolute error (MAE) of 3.56 mmHg for systolic blood pressure (SBP), outperforming existing non-contact approaches. Our results demonstrate strong potential for real-time, unobtrusive BP monitoring, especially in applications such as remote health assessment, smart environments, and wearable healthcare systems. Our findings also highlight the utility of chaos modeling as a viable tool for capturing the complex, nonlinear nature of cardiovascular dynamics in a non-invasive manner.