Hypertension is a major contributor to global mortality, necessitating continuous and noninvasive BP monitoring to manage its impact. This work introduces UPR-BP, an unsupervised learning framework that leverages unlabeled PPG signals for accurate BP estimation, addressing key challenges in conventional cuff-based and data-driven models. By utilizing massive unlabeled PPG data, UPR-BP achieves medical-grade accuracy in BP estimation with minimal labeled data. The model leverages advanced techniques like temporal neighborhood coding and VICReg, optimizing feature representations of PPG signals to capture subtle physiological changes. Compared to existing models that often rely on additional signals like ECG or invasive calibration, UPR-BP is more scalable and practical for continuous, long-term BP monitoring in clinical settings. This leap in methodology and application demonstrates potential for enhancing real-world BP monitoring accuracy and accessibility.

This study presents unsupervised photoplethysmography representation learning for noninvasive blood pressure (UPR-BP), a novel framework utilizing photoplethysmography (PPG) signals for accurate, noninvasive BP estimation. Leveraging readily available unlabeled PPG data, UPR-BP overcomes the limitations of data-driven models by effectively capturing discriminative BP features without requiring extensive paired measurements. Our framework employs a three-branch architecture with shared weights for joint optimization and incorporates preservation of invariance, variance, and covariance in the PPG temporal encoding, preventing information collapse and generating meaningful deep representations. Additionally, temporal neighborhood coding facilitates the identification of diverse physiological states within the PPG signals. We comprehensively validate UPR-BP on diverse datasets from bedside monitors and wearable wristwatches, encompassing over 4000 subjects. The proposed approach achieves m...