A Scoping Review of Deep Learning Methods for Photoplethysmography Data

Summary

This comprehensive scoping review, led by Guangkun Nie and Shenda Hong's group at Peking University, synthesises 193+ papers applying deep learning to PPG signals published between 2017 and 2025. The review is structured around three axes: tasks (what health signals are extracted), models (which architectures are used), and data (which datasets and benchmarks are used). On the task side, blood pressure analysis dominates (72 papers), followed by cardiovascular monitoring and diagnosis (43), signal quality assessment, sleep health, mental health, and non-medical tasks.