Intelligent Healthcare Monitoring System using Machine Learning

Abstract

The growing need for affordable and intelligent healthcare support has encouraged the integration of Internet of Things (IoT) devices with machine learning-based analysis. Conventional healthcare systems generally rely on periodic clinical visits, which can delay the identification of abnormal physiological patterns. This work presents an intelligent realtime healthcare monitoring and screening system that combines live sensor-based data acquisition, ensemble machine learning, and a web-based monitoring interface. The system uses manually entered parameters such as age, gender, and body mass index (BMI), along with sensor-collected physiological values including body temperature, heart rate, and $\mathbf{S p O}_{\mathbf{2}}$. The hardware implementation is based on Arduino UNO R4 WiFi and ESP32-S3 integrated with MAX30102, MLX90614, and DHT22 sensors. The measured data is transmitted to a backend server where an ensemble model consisting of Logistic Regression, Random Forest, and XGBoost performs real-time prediction. The proposed ensemble model achieved an accuracy of 93.62%, while the ANN deep learning baseline achieved 95.74% on the same test dataset. The ensemble model remains suitable for the proposed IoT-based screening system because it provides competitive performance, stable prediction behaviour, and better interpretability than black-box deep learning models. The system is positioned as a screening and monitoring tool rather than a clinical diagnosis system. It generates interpretable health condition outputs such as Physiologically Stable, Mild Physiological Risk, Moderate Physiological Screening Risk, and High Physiological Screening Risk. The results show that the proposed framework effectively combines low-cost IoT hardware and machine learning for real-time health monitoring through a user-friendly interface.