Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Deep Imbalanced Regression to Estimate Vascular Age from PPG Data: a Novel Digital Biomarker for Cardiovascular Health

Guangkun Nie, Qinghao Zhao, Gongzheng Tang, Jun Li, Shenda Hong

TL;DR

The study tackles the bias caused by imbalanced age distributions in predicting vascular age from PPG signals. It introduces Dist loss, a distribution‑aware loss term, combined with a standard regression loss as $L_{total}=M(\hat{Y},Y)+\lambda L_{dist}$ and $L_{dist}=M(\hat{Y}_{sort},Y^{E}_{sort})$, where $Y^{E}$ is derived from a KDE‑estimated label distribution. Evaluations on the UK Biobank (n=502,389) show state‑of‑the‑art performance, particularly in few‑shot regions, and clinical validation demonstrates that predicted vascular age associates with mortality, CHD, heart failure, and arterial stiffness. The results support vascular age from PPG as a robust digital biomarker for cardiovascular health and risk stratification in large populations.

Abstract

Photoplethysmography (PPG) is emerging as a crucial tool for monitoring human hemodynamics, with recent studies highlighting its potential in assessing vascular aging through deep learning. However, real-world age distributions are often imbalanced, posing significant challenges for deep learning models. In this paper, we introduce a novel, simple, and effective loss function named the Dist Loss to address deep imbalanced regression tasks. We trained a one-dimensional convolutional neural network (Net1D) incorporating the Dist Loss on the extensive UK Biobank dataset (n=502,389) to estimate vascular age from PPG signals and validate its efficacy in characterizing cardiovascular health. The model's performance was validated on a 40% held-out test set, achieving state-of-the-art results, especially in regions with small sample sizes. Furthermore, we divided the population into three subgroups based on the difference between predicted vascular age and chronological age: less than -10 years, between -10 and 10 years, and greater than 10 years. We analyzed the relationship between predicted vascular age and several cardiovascular events over a follow-up period of up to 10 years, including death, coronary heart disease, and heart failure. Our results indicate that the predicted vascular age has significant potential to reflect an individual's cardiovascular health status. Our code will be available at https://github.com/Ngk03/AI-vascular-age.

Deep Imbalanced Regression to Estimate Vascular Age from PPG Data: a Novel Digital Biomarker for Cardiovascular Health

TL;DR

The study tackles the bias caused by imbalanced age distributions in predicting vascular age from PPG signals. It introduces Dist loss, a distribution‑aware loss term, combined with a standard regression loss as and , where is derived from a KDE‑estimated label distribution. Evaluations on the UK Biobank (n=502,389) show state‑of‑the‑art performance, particularly in few‑shot regions, and clinical validation demonstrates that predicted vascular age associates with mortality, CHD, heart failure, and arterial stiffness. The results support vascular age from PPG as a robust digital biomarker for cardiovascular health and risk stratification in large populations.

Abstract

Photoplethysmography (PPG) is emerging as a crucial tool for monitoring human hemodynamics, with recent studies highlighting its potential in assessing vascular aging through deep learning. However, real-world age distributions are often imbalanced, posing significant challenges for deep learning models. In this paper, we introduce a novel, simple, and effective loss function named the Dist Loss to address deep imbalanced regression tasks. We trained a one-dimensional convolutional neural network (Net1D) incorporating the Dist Loss on the extensive UK Biobank dataset (n=502,389) to estimate vascular age from PPG signals and validate its efficacy in characterizing cardiovascular health. The model's performance was validated on a 40% held-out test set, achieving state-of-the-art results, especially in regions with small sample sizes. Furthermore, we divided the population into three subgroups based on the difference between predicted vascular age and chronological age: less than -10 years, between -10 and 10 years, and greater than 10 years. We analyzed the relationship between predicted vascular age and several cardiovascular events over a follow-up period of up to 10 years, including death, coronary heart disease, and heart failure. Our results indicate that the predicted vascular age has significant potential to reflect an individual's cardiovascular health status. Our code will be available at https://github.com/Ngk03/AI-vascular-age.
Paper Structure (27 sections, 10 equations, 4 figures, 4 tables)

This paper contains 27 sections, 10 equations, 4 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Methods
  3. Problem definition
  4. Model architecture
  5. Deep imbalanced regression
  6. Overview
  7. The Dist loss
  8. Experiments and Results
  9. Experimental setup
  10. Dataset and preprocessing
  11. Evaluation metrics
  12. Pearson Correlation Coefficient
  13. MAE & RMSE
  14. Weighted MAE & RMSE
  15. Overlap ratio
  16. ...and 12 more sections

Figures (4)

  • Figure 1: Workflow of our vascular age estimation method.
  • Figure 2: Comparison between the predicted vascular age by the plain model (a, b) and our method (c, d), $r$ denotes Pearson correlation coefficient. (a) Illustrates the distribution of chronological age (light blue bars) and the corresponding vascular age predicted by the plain model (orange bars) in the UKBB dataset. (b) Presents a scatter plot highlighting the disparity between chronological age and predicted vascular age by the plain model. (c) Depicts the distribution of chronological age (light blue bars) and the corresponding vascular age predicted by our method (orange bars). (d) Displays a scatter plot showcasing the higher consistency between chronological age and predicted vascular age achieved by our method.
  • Figure 3: Scatter plots illustrating the relationship between arterial stiffness and chronological age (a) and predicted vascular age (b). $r$ denotes Pearson correlation coefficient.
  • Figure 4: Adjusted KM curves for different outcomes, including death, CHD, and heart failure, by sex, age, and ethnic background.