Machine Learning-Driven IoT System for Patient Monitoring

Vollständiger Abstract

Worum geht es in dieser Arbeit?

In this paper, a healthcare monitoring system is introduced under IoT based on WiFi enabled Arduino boards to deliver patient data to the cloud through the ThingSpeak platform. It checks abnormalities in health, such as body temperature, level of oxygen, heart rate and vibration, which are checked by the system. A Long Short-Term Memory (LSTM) model is used to generate a synthetic dataset consisting of 14,664 samples, each labeled as either normal or abnormal based on expert assessment. A comparative approach between the traditional machine learning method and the hybrid machine learning model is introduced in this paper. The results demonstrate the effectiveness of the hybrid machine learning model in detecting abnormal patient conditions in the traditional machine learning approach, the PCA-SVM (Principal Component Analysis Support Vector Machine) model (achieved an accuracy of 98.87%, while the Random Forest model yielded the highest accuracy at 99%. The integration of hybrid machine learning methods achieved an overall accuracy of approximately 99%. Beyond technical performance, this work demonstrates the feasibility of deploying low-cost, scalable IoT systems for continuous patient monitoring in both clinical and home settings. The findings underscore the potential of hybrid machine learning architectures to enhance the reliability and accuracy of automated health diagnostics, contributing to proactive healthcare delivery and supporting the development of more responsive remote patient management systems.