Non-invasive monitoring of human vital signs under dynamic conditions: A review of acoustic, optical and ballistocardiographic methods

Introduction: Modern healthcare systems face a critical shortage of personnel, which in the traditional model leads to dangerous “time windows” (reaching up to 22 h per day) during which the patient remains unsupervised. The response to this challenge is the dynamic development of contactless monitoring technologies, enabling continuous assessment of vital parameters without the discomfort associated with cabling. However, the implementation of these solutions in real-world environments encounters a fundamental barrier in the form of motion artifacts, variable environmental conditions, and demographic heterogeneity. Aim: This article provides a review of non-invasive monitoring methods under dynamic conditions, with a particular emphasis on interference compensation techniques. The aim of the work is a comparative analysis of three leading modalities: optical and guided-wave photonic (rPPG, thermography, LiDAR, FBG, FPI), vibrational-mechanical (BCG, SCG), and acoustic (passive and active), in the context of their clinical reliability, privacy protection, and motion robustness. Methodology: The review includes a literature analysis from the years 1947–2025, with a special emphasis on the “digital era” after 2010. The effectiveness of modern signal processing algorithms was analyzed, including bounded Kalman filters, Tile and Aggregation (T&A) methods for diverse skin pigmentation, and deep learning architectures (UNet, Bi-LSTM) used for the reconstruction of biomedical signals, alongside targeted strategies evaluated for specific vulnerable demographics (high BMI, preterm infants, and elderly populations). Results: The analysis shows a clear specialization of individual technologies. Optical methods (rPPG) offer the highest precision in measuring heart rate and saturation, particularly in neonatology (SIDS prevention), but are limited by lighting conditions and privacy issues, whereas guided-wave photonics offer complete electromagnetic immunity. Vibrational methods (BCG/SCG) provide discrete “zero-effort” measurement and unique hemodynamic data (PEP, LVET), yet they exhibit critical sensitivity to patient motor activity. Acoustic technologies, especially active ultrasonic sonars, represent a compromise that allows for respiratory monitoring in the dark and through obstacles. Conclusions: There is currently no single, universal measurement method robust to all environmental interferences. The future of non-invasive monitoring lies in hybrid systems (sensor fusion), combining the advantages of different modalities (e.g., optics and vibration), and in edge processing (Edge AI). Addressing the strict trade-offs between computational overhead in matrix-inversion algorithms and hardware power constraints will be key. Only the integration of multi-source data will allow for the effective elimination of motion artifacts and the implementation of medical safety-critical systems in home and mobile conditions.