Estimating Heart Rate from Inertial Sensors Embedded in Smart Eyewear: A Validation Study

Highlights: What are the main findings? The “Essilor Connected Glasses”, using an IMU for heart rate estimation, showed high accuracy compared to ECG and smartwatch during controlled static activities. The quality index (QI) significantly impacted heart rate estimation accuracy, with higher QI values correlating with lower error rates. What is the implication of the main finding? Smart eyewear with IMUs offers a viable, non-invasive alternative for opportunistic and unobtrusive heart rate monitoring during sedentary activities. QI thresholds can be used to optimize wearable HR monitoring systems for improved performance in different physiological states. Smart glasses are promising alternatives for the continuous, unobtrusive monitoring of heart rate (HR). This study validates HR estimates obtained with the “Essilor Connected Glasses” (SmartEW) during sedentary activities. Thirty participants wore the SmartEW, equipped with an IMU sensor for HR estimation, a commercial smartwatch (Garmin Venu 3), and an ECG device (Movesense Flash). The protocol included six static tasks performed under controlled laboratory conditions. The SmartEW algorithm analyzed 22.5 s signal windows using spectral analysis to estimate HR and provide a quality index (QI). Statistical analyses assessed agreement with ECG and the impact of QI on HR accuracy. SmartEW showed high agreement with ECG, especially with QI threshold equal to 70, as a trade-off between accuracy, low error, and acceptable data coverage (80%). Correlation for QI ≥ 70 was high across all the experimental phases (r2 up to 0.96), and the accuracy within ±5 bpm reached 95%. QI ≥ 70 also allowed biases to decrease (e.g., from −1.83 to −0.19 bpm while standing), with narrower limits of agreement, compared to ECG. SmartEW showed promising HR accuracy across sedentary activities, yielding high correlation and strong agreement with ECG and Garmin. SmartEW appears suitable for HR monitoring in static conditions, particularly when data quality is ensured.