Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
August 6, 2026Physiological MeasurementOpen Access

Practical Optimization of Deep Learning Models for Cuffless Blood Pressure Estimation From Photoplethysmography

View Full Paper
Ask AI
Bookmark
Share

Authors

VKVikas KhuranaMPMohammad Aejaz PamidiBJBharat Bhushan Joshi Bhushan Joshi

Discussion

Loading...

Member takes

Overview

Randomized trial evaluates cuffless blood pressure estimation in intensive care, suggesting optimal methods for continuous monitoring.

Key Points

  • The aim is to enhance cuffless blood pressure estimation using deep learning from photoplethysmography data.
  • Analyzed 205,850 paired PPG-arterial BP segments from the MIMIC waveform database.
  • Compared sampling frequencies of 125, 62.5, and 31.25 Hz using a MultiResUNet model.
  • Calculated SBP and DBP via minimum-maximum extraction or averaging peaks and troughs.
  • 62.5 Hz input reduced MAE for SBP (5.24 vs 6.34 mmHg) and DBP (2.87 vs 3.27 mmHg) compared to 125 Hz.
  • 31.25 Hz increased MAE significantly for SBP (16.09 mmHg) and DBP (9.10 mmHg).
  • Lower MAE was associated with moderate respiratory rates, while low and high SBP ranges showed higher errors.

Cite This Study

Khurana et al. (2026) studied this question.

synapsesocial.com/papers/6a743730764cddc9499d4790https://doi.org/10.1088/1361-6579/ae944f
View Full Paper
Ask AI
Bookmark
Share