Adaptive Baseline Modeling for Personalized Cardiovascular Anomaly Detection

Abstract

Wearable devices such as smartwatches allow continuous monitoring of physiological signals, including heart rate and activity levels. Many current monitoring systems rely on fixed population-based thresholds that may not reflect individual physiological differences. This paper explores a personalised monitoring framework based on adaptive baseline modelling and anomaly detection. Physiological signals obtained from wearable sensors such as photoplethysmography (PPG), accelerometers, and gyroscopes are used to derive features including heart rate, heart rate variability, and motion activity. By learning an individual’s normal physiological patterns over time, the system can identify deviations that may indicate unusual cardiovascular behaviour. The goal of this work is to outline a monitoring approach to support personalised health monitoring using wearable devices.