Challenges And Prospects Of Visual Contactless Physiological Monitoring In Clinical Study

This article discusses the potential of Vital-Signs-from-Video-based Physiological Measurement (VCPM) technology and its future directions. It is organized into four sections, with the first section discussing existing relevant works and study characteristics. The second section discusses the revolution of digital medicine, the merits of VCPM technology and the necessity of clinical settings. The third section illustrates the main challenges in clinical studies, and the fourth section presents the future directions and prospects of VCPM. These future directions are summarized from three perspectives: the adoption of SOTA deep learning technologies, the breakthrough in the current limitations and clinical application challenges, and the establishment of a national and international standard of VCPM system. Unsupervised learning techniques and Federated Learning (FL) are discussed as potential solutions.