A Self-Calibrated, Wearable Vital Sign Monitoring Device Enabled by Edge Computing

The application of wearable technology in healthcare is revolutionizing personalized and proactive medical care. Continuous vital sign monitoring plays a crucial role in early disease diagnosis and prevention; however, many existing wearable solutions face challenges such as poor sensor signal quality and dependency on cloud connectivity, which raises data security and latency issues. This study presents a flexible wearable sensing device equipped with four MAX30101 photoplethysmogram (PPG) sensors and an edge computer utilizing a pre-trained Bayesian optimization model for self-calibration. Unlike conventional devices, the proposed system performs real-time, on-device processing, eliminating reliance on external networks and enhancing user privacy and responsiveness. Optimal sensor parameters are identified, ensuring high-quality pulse waveforms, with 90% of the data achieving optimal calibration. HR and RR are directly extracted from PPG signals, while BP is estimated from pulse wave velocity and other predictors (age, gender, HR, weight, and height) using a multilayer perceptron model. Testing on 30 adults (aged 17–50) with varying skin tones demonstrated average differences of 2.21 BPM for HR and ±1.35 BrPM for RR, with systolic and diastolic BP predictions achieving R² scores of 0.95 and 0.90, respectively. The device autonomously calibrates within 60 seconds, enabling accurate, real-time vital sign monitoring. By addressing key limitations in existing wearables, this solution not only facilitates early diagnosis and improved cardiovascular health management but also paves the way for next-generation, privacy-preserving healthcare technologies.