Multi-objective Feature Selection in Remote Health Monitoring Applications

TL;DR

This paper addresses privacy-aware remote health monitoring by balancing breathing pattern recognition against user identification. It formulates a multi-objective optimization with $O_1 = a_R$, $O_2 = 1 - a_I$, and $O_3 = a_R - a_I$, solved via NSGA-II to select feature subsets from $F$. Features are extracted from mmWave radar signals yielding 189 features; two RF-based classifiers are trained on selected subsets to measure trade-offs. On a dataset of 50 subjects performing four breathing patterns in two positions, the approach achieves a disparity in accuracy (breathing recognition around ~86% vs user identification around ~30%), and surrogate fitness via PCA+k-NN reduces computation by roughly a factor of 3, illustrating practical, privacy-aware remote health monitoring.

Abstract

Radio frequency (RF) signals have facilitated the development of non-contact human monitoring tasks, such as vital signs measurement, activity recognition, and user identification. In some specific scenarios, an RF signal analysis framework may prioritize the performance of one task over that of others. In response to this requirement, we employ a multi-objective optimization approach inspired by biological principles to select discriminative features that enhance the accuracy of breathing patterns recognition while simultaneously impeding the identification of individual users. This approach is validated using a novel vital signs dataset consisting of 50 subjects engaged in four distinct breathing patterns. Our findings indicate a remarkable result: a substantial divergence in accuracy between breathing recognition and user identification. As a complementary viewpoint, we present a contrariwise result to maximize user identification accuracy and minimize the system's capacity for breathing activity recognition.

Figures (14)

• Figure 1: Feature importance in two tasks
• Figure 2: Multi-objective bio-inspired feature selection
• Figure 3: Four different breathing activities captured with an mmWave radar.
• Figure 4: Detected peaks in the breathing signals: we extract Respiratory Rate Variability using the detected peaks.
• Figure 5: User positions in the data collection sessions