Title: Robust heart rate estimation during intensive physical training using AI-enhanced particle filter

Authors: Kokila Bharti Jaiswal

Addresses: Department of ECE, Bhilai Institute of Technology, Durg, 491001, India

Abstract: The remote photoplethysmography (rPPG) signal provides an essential data for the estimation of heart rate (HR). However, rPPG signal is often corrupted by noise due to motion, driving the conventional denoising techniques to failure. Perhaps, the motion artefact is of more concern when it is falsely captured as a real pulse signal. In this article, a novel methodology is proposed leveraging particle filters (PF) for the robust HR measurement in the presence of motion artefact. The proposed method improves the measurement accuracy of HR specifically for the cases of physical exercise, when the subject is severely corrupted by motion artefacts. Proposed approach yields better results in terms of estimation of HR on benchmark datasets such as UBFC-rPPG and PURE. Moreover, the proposed method is a stand-alone technique, and can be easily associated with the existing algorithms.

Keywords: remote health monitoring; heart rate; remote photoplethysmography; rPPG; particle filter; denoising; motion artefact.

DOI: 10.1504/IJAIH.2025.149254

International Journal of Artificial Intelligence in Healthcare, 2025 Vol.1 No.1, pp.92 - 108

Received: 08 Apr 2025
Accepted: 10 Aug 2025
Published online: 20 Oct 2025 *