A Data-Driven Framework for Stress Detection from Wearable Sensors in Smart Healthcare

Abstract

With the development of technology and increasing demands, the integration of Internet of Things (IoT) and wearable sensors into urban infrastructure including smart healthcare becomes a pillar of smart cities. This work presents a data-driven framework for stress level predicting by using physiological and behavioral data collected from wearable devices. The methodology proposed follows four key stages: (1) data preprocessing, (2) feature engineering, (3) model development and (4) metrics evaluation. Also, four machine learning (ML) algorithms are trained and tested on a multi-modal dataset including: decision tree (DT), random forest (RF), light gradient boosting machine (LightGBM), and Categorical Boosting algorithm (CatBoost). The results show that boosting algorithms, particularly LightGBM and CatBoost, outperformed the traditional DT and achieved the highest ROC-AUC values (0.6354 and 0.6326) respectively. This study underlines the importance of integrating wearable technologies and ML in order to enhance preventive healthcare within smart cities.