Original scientific paper

https://doi.org/10.17559/TV-20240828001950

Performance Evaluation of Wrist Pulse Signals by using Hybrid Artificial Bee Colony with Feed Forward Neural Networks

V. R. Vijaykumar ; Department of Electronics Engineering, Anna University (MIT Campus), Chrompet, Chennai, India
K. Vijayagopal ; Department of EEE, Anna University, Chennai, India *

* Corresponding author.

Abstract

People are increasingly suffering from various health issues, with a noticeable rise in frequency. This has created a substantial global demand for health assessments, as these problems are affecting all socioeconomic groups more widely. Technological advancements in disease prevention and health maintenance have led to the development of new fields, such as monitoring systems. Heart rate, which is the average number of times the heart beats per minute, reflects various physical states, including workload, stress, attention levels, drowsiness, and the activity of the nervous system. Deployment of IoT based pulse sensor devices is a more effective way of detecting pulse signal from wrist that is termed wrist pulse signal. Using the SN11574 model pulse sensor, real-time signals from both healthy and unhealthy individuals can be collected with the help of an Arduino controller. The captured signals are processed such as noise removal, signal normalization and signal quantization etc. The sensor's analog pulse rate readings are converted into digital data, which is then analyzed using the Fast Fourier Transform (FFT). This method has an advantage over traditional FFT approaches in reducing the influence of breathing patterns and avoiding the mixing of breathing and heartbeat signals. The classification of pulse signals for healthy and unhealthy individuals is done using a combination of the Hybrid Artificial Bee Colony and a Feed Forward Neural Network (HABCFFNN) to predict the individual's health status.The pulse databases are implemented in this HABCFFNN for performance evaluation. The wrist pulse signal databases, POLYU and Wojcikowski, are used in this method. The output of the neural network is fed into an analysis report, where some statistical notations are considered. Performance measurements such as accuracy, precision, sensitivity, specificity, etc., are used to highlight the effectiveness of this method compared to other related work.

Keywords

disease prevention technology; health monitoring; heart rate detection; neural network classification; pulse signal analysis

Hrčak ID:

335047

URI

https://hrcak.srce.hr/335047

Publication date:

30.8.2025.

Visits: 334 *