Multitask Learning for PPG Applications: Signal Quality Assessment and Physiological Parameter Estimation

This project advances wearable health monitoring by applying multitask learning (MTL) to photoplethysmography (PPG) signals. Instead of training separate models for each parameter, we design deep learning models that jointly learn related tasks, such as PPG quality assessment, heart rate (HR), heart rate variability (HRV), and respiration rate (RR) estimation. By leveraging shared physiological characteristics, the MTL framework improves accuracy in assessing signal quality and reduces error rates in HR and RR estimation. Evaluated on smartwatch data from 46 participants in daily life, the approach demonstrates how multitask models enhance both reliability and efficiency in real-world PPG-based applications.

Publications:


Full paper

Project information

  • Category: Future Healthcare
  • Contact Person: Amir M. Rahmani
  • Multitask Learning for PPG Applications: Signal Quality Assessment and Physiological Parameter Estimation

info@futurehealth.uci.edu

© Copyright 2021 UCI Institute for Future Health - All Rights Reserved