Anomaly Detection in IoT Healthcare Security using Machine Learning Methods

Razan Alharith, Hiba Ahmed, Ashraf Osman Ibrahim, Mohammed A. Saleh, Amanzholova Saule Toksanovna, Adilzhanova Saltanat

2025

Closed access Metadata (CSL-JSON)
DOI 10.1109/sist61657.2025.11139151
Data confidence 70 medium
How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

The incorporation of the Internet of Things (IoT) within the healthcare sector significantly improves operational efficiency, elevates patient care, and increases cost-effectiveness. Nonetheless, it introduces obstacles including data security, interoperability, and ethical considerations. Devices connected through the Internet of Things gather health data from patients and send it to servers. Here, machine learning algorithms analyze the information to identify abnormal vital signs and enable early detection of diseases. Considering the vulnerability of these devices to cyber threats, implementing robust security measures is essential. The study utilizes the WUSTL EHMS 2020 dataset, integrating patient biometrics with network flow data, to create machine learning algorithms aimed at detecting atypical network traffic associated with IoT medical devices. The dataset is subjected to preprocessing to establish balanced supervised models, which encompass Support Vector Machine (SVM), Naive Bayes, Random Forest and Logistic Regression. The Random Forest model demonstrated an impressive 100% accuracy and an ideal F1 score, establishing it as the optimal selection for real-time IoT security applications. Naive Bayes and SVM demonstrated commendable performance, yet their accuracy and recall rates fell short. Logistic Regression, although beneficial, encountered challenges with false positives and false negatives. The results highlight the significance of choosing the right model in IoT security, with Random Forest emerging as the most dependable approach for ensuring consistent threat detection.