What question did this study set out to answer?

The aim is to develop a secure framework for continuous health monitoring that protects sensitive data.

March 4, 2026Open Access

IoT Based Security Enhanced Continuous Health Monitoring Framework Using Lightweight Cryptography and Deep Learning Techniques

Puntos clave

The aim is to develop a secure framework for continuous health monitoring that protects sensitive data.
Integration of lightweight cryptography with deep learning techniques.
Health data is encrypted using a chaos-based cryptographic algorithm.
Feature selection is performed using a modified tomtit flock optimization algorithm.
Data analysis utilizes a pre-trained DenseNet model.
Predictions are made with a deep Bayesian spatial-temporal graph neural network (DBST-GNN).
The framework achieved a classification accuracy of 99.416% on the real-time student dataset.
The model reached a classification accuracy of 99.625% on the Kaggle Elikplim dataset.
Validated using real-time student health data to demonstrate effectiveness in monitoring.

Resumen

The rapid growth of networks and the Internet of Things (IoT) has increased the speed of establishing behavioral and physiological data management systems, specifically for continuous health monitoring.Recent advances in wearables devices and mobile health apps allow real-time tracking of student’s vital signs and detects biological and behavioral changes via network-based data collection and AI-related analytics.Although such systems are very useful particularly to physically active students they also raise critical issues regarding safe data collection, transmission, and protection of privacy. To tackle these issues, this paper proposes a continuous health monitoring Framework to secure the IoT enabled systems by integrating lightweight cryptography with deep learning techniques. In the process of data transmission, sensitive health data is encrypted using chaos-based lightweight cryptographic algorithm, and the signature of the encryption key is securely stored in the cloud to inhibit any unauthorized access.To perform the data analysis, a pre-trained DenseNet architecture is used to obtain discriminative deep features of collected health traces. In order to maximize efficiency, modified tomtit flock optimization (MTFO) algorithm is used to select features to reduce the dimensionality of data and preserve the most important information.A deep Bayesian spatial-temporal graph neural network (DBST-GNN) then predicts the prognosis bycontinuously monitoring the student’s health based on the biological and social health factors.The proposed framework is validated using a real-time student health dataset and the Kaggle Elikplim dataset to demonstrate the impact of secure data sharing and artificial intelligence analysis on student health monitoring, personalized counseling, and overall health improvement. Experimental results show that the proposed model achieves high classification accuracy of 99.416% on the real-time student dataset and 99.625% on Kaggle Elikplim dataset.

Me gusta

Guardar

Ver artículo completo