Innovative Sensing Approaches for Real-Time Physiological Monitoring and Stress Evaluation

The conducted study investigates advanced methods for non-invasive physiological data acquisition and their application in stress assessment. The research explores the potential of the Nanomade Pulse sensor, a novel high-sensitivity strain gauge technology, for monitoring cardiac (Inter-Beat Interval, IBI or Heart Rate, HR), respiratory (Inter-Respiratory Interval, IRI or Respiratory Rate, RR), and movement (Actimetry, ACT) parameters. The initial phase involved validating the Nanomade Pulse sensor demo-kit against reference devices (Polysomnography, Garmin Venu 2 Plus, and Polar H10) in a controlled study with 11 participants (9 females and 2 males), aged 24 ± 3 years, under different postural and metabolic conditions. Although the sensor demonstrated promising capabilities, the data quality proved not even consistent for accurate Heart Rate Variability (HRV) analysis, highlighting the current technological limitations of the sensor demo-kit. In terms of results, the IBI measurement accuracy, quantified by the Root Mean Square Error (RMSE), varied significantly, ranging from 150ms to 2500ms, with a mean RMSE of 1120ms. To ensure robust stress evaluation, the study subsequently leveraged high-quality HRV and physiological data from Garmin and Polar devices. A second experimental phase with 31 participants employed an incremental cycling test to induce progressive cardiovascular load and stress. Objective HRV indicators (from Polar H10) and subjective Ratings of Perceived Exertion (RPE) were combined to characterize physiological and perceived stress responses. A comprehensive analysis framework was implemented, encompassing time-domain (STD, RMSE, RMSSD, pNN50), frequency-domain (Lomb–Scargle Periodogram, LF, HF, LF/HF ratio), and non-linear (Sample Entropy) metrics. A weighted Stress Score was then formulated—integrating RMSSD, LF/HF ratio, Mean Heart Rate, and High-Frequency Normalized Units (HFnu)—and normalized on a 1–10 scale for comparative evaluation with other metrics. Crucially, this score showed a strong, statistically significant correlation with the subjective Ratings of Perceived Exertion (RPE) (r = 0.88, p < 0.001). The Stress Score progression was highly consistent with the incremental protocol, demonstrating a continuous increase along the test phases, reflecting the transition from low to maximum cardiovascular load. Efficacy was further demonstrated in an exploratory single-subject continuous monitoring analysis (Garmin data), validating the model's potential for dynamic stress tracking in everyday life. Ultimately, this thesis bridges the gap between signal quality and stress assessment accuracy by investigating next-generation non-invasive sensing technologies. The findings lay the foundation for future development of reliable ambient systems for continuous physiological and stress monitoring in real-world environments, like driving contexts.