Implementation of an Android-embedded BIA system with machine learning-ready signal processing for precision body composition analysis

Abstract

This paper presents an advanced Android-based embedded system for body fat measurement using Bioelectrical Impedance Analysis (BIA). The proposed system features an integrated pipeline of signal generation, adaptive acquisition, and real-time processing within a single embedded platform, significantly enhancing portability for smart healthcare manufacturing. A micro alternating current ($50 text{ kHz}, 100 text{ mu A}$) is applied to the human body, and the resulting impedance signals are stabilized using optimized low-pass and high-pass filtering techniques to ensure data fidelity for subsequent analysis. Beyond conventional empirical equations, this platform is designed to support Machine Learning (ML)-based estimation models by providing high-resolution digital impedance data. Experimental results confirm stable signal acquisition and reliable real-time visualization through a dedicated Android application. This integrated approach offers a scalable solution for intelligent point-of-care devices in the era of Artificial Intelligence of Things (AIoT).