Application of Real-Time Palm Imaging with Nelder-Mead Particle Swarm Optimization/Regression Algorithms for...

Area of Science:

• Biomedical Engineering
• Health Informatics
• Medical Devices

Background:

• Hypertension prevalence is increasing due to lifestyle changes, necessitating accessible monitoring.
• Hypertension is a "silent killer" requiring early and convenient detection methods.
• Traditional blood pressure (BP) monitoring devices are often invasive and less accessible for home use.

Purpose of the Study:

• To develop and compare two novel non-contact methods for blood pressure (BP) monitoring.
• To assess the accuracy and convenience of these methods for home-based hypertension management.
• To explore cost-effective, non-invasive alternatives to traditional BP measurement devices.

Main Methods:

• A Nelder-Mead simplex with particle swarm optimization (NM-PSO) approach using webcam palm imaging and Independent Component Analysis (ICA) for noise reduction.
• A machine learning regression analysis method utilizing derived physiological data for BP estimation.
• Both methods extract data from palm images captured via a standard webcam.

Main Results:

• The NM-PSO method achieved a root mean square error (RMSE) of 2.71 mmHg for systolic BP (SBP) and 3.42 mmHg for diastolic BP (DBP).
• The machine learning regression method yielded an RMSE of 2.88 mmHg for SBP and 2.60 mmHg for DBP.
• Both methods provided accurate BP measurements within 10 seconds, demonstrating suitability for home use.

Conclusions:

• Both NM-PSO and machine learning regression offer fast, accurate, and convenient non-contact BP monitoring.
• The NM-PSO method excels in noise handling, while regression offers formulaic efficiency.
• These biomimetic approaches present a viable, cost-effective alternative to traditional contact-based BP monitors, enhancing hypertension management accessibility.