Scholarlink

Hypertension, a pervasive cardiovascular condition affecting a significant portion of the global populace, elevates the risk of various cardiovascular events and complications. Monitoring elevated blood pressure levels in individuals, especially in those predisposed to cardiovascular issues, via accessible wearables can herald significant advancements in preventive healthcare. As wearable technology becomes more cost-effective, there is an increasing emphasis on tailored healthcare solutions, emphasizing the importance of on-device bio-signal processing to mitigate challenges related to centralized data processing, such as latency and network vulnerabilities. This study utilizes the MIMIC II dataset and meets the requirements specified by many medical institutions by achieving mean absolute error (MAE) of 4.23 mmHg for systolic blood pressure (SBP) and 2.58 mmHg for diastolic blood pressure (DBP) estimation using Bonsai, an Edge ML model. The extensive feature engineering process, coupled with edge model deployment, ensures accurate and reliable blood pressure monitoring, contributing to the development of effective, real-time healthcare solutions.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Blood Pressure Estimation from Photoplethysmography Signals: An EdgeML Approach

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • P. Prithvikiran,
  • V. Varun Vijay Kumar,
  • Nithin Aditya,
  • R. Aishwarya,
  • S. Mohanavalli

摘要

Hypertension, a pervasive cardiovascular condition affecting a significant portion of the global populace, elevates the risk of various cardiovascular events and complications. Monitoring elevated blood pressure levels in individuals, especially in those predisposed to cardiovascular issues, via accessible wearables can herald significant advancements in preventive healthcare. As wearable technology becomes more cost-effective, there is an increasing emphasis on tailored healthcare solutions, emphasizing the importance of on-device bio-signal processing to mitigate challenges related to centralized data processing, such as latency and network vulnerabilities. This study utilizes the MIMIC II dataset and meets the requirements specified by many medical institutions by achieving mean absolute error (MAE) of 4.23 mmHg for systolic blood pressure (SBP) and 2.58 mmHg for diastolic blood pressure (DBP) estimation using Bonsai, an Edge ML model. The extensive feature engineering process, coupled with edge model deployment, ensures accurate and reliable blood pressure monitoring, contributing to the development of effective, real-time healthcare solutions.