A Secure Framework for MIoT: TinyML-powered Emergency Alerts and Intrusion Detection for Secure Real-time Monitoring

The healthcare industry is shifting towards a patient-centered model, with remote monitoring at its core. The Medical Internet of Things (MIoT) is central to this shift, leveraging sensors to collect real-time health data and enhance patient care. However, managing sensitive information raises privacy concerns, making robust data security crucial for system success. Our proposed solution addresses these challenges by using edge processing to encrypt personal data before it is sent to the cloud, and applying Tiny Machine Learning (TinyML) for real-time vital sign analysis to trigger emergency alerts. Additionally, we incorporate a lightweight Intrusion Detection System (IDS) to identify potential attacks. The proposed system is validated in two phases: heart disease prediction using TinyML achieved 88% accuracy with Random Forest. The sensitive data were encrypted using a lightweight AES scheme in Edge before being in the cloud. In the second phase, the Intrusion Detection System (IDS) using IoT Flock data demonstrated 99.67% accuracy with K-Nearest Neighbors (KNN) and a low false positive. This approach enhances real-time health monitoring efficiency and data security, marking a notable advancement in MIoT applications.