Anomaly detection in IoT-based healthcare: machine learning for enhanced security.

Area of Science:

• Cybersecurity
• Machine Learning
• Healthcare Technology

Background:

• Internet of Things (IoT) integration in healthcare offers significant benefits but faces challenges in data security and privacy.
• IoT devices collect sensitive patient data, making them vulnerable to cyberattacks.
• Effective detection of anomalous network traffic is crucial for securing healthcare IoT systems.

Purpose of the Study:

• To develop and evaluate machine learning models for efficient detection of anomalous network traffic in healthcare IoT environments.
• To assess the performance of various supervised learning algorithms using the Canadian Institute for Cybersecurity (CIC) IoT dataset.
• To optimize models for real-time attack detection and response.

Main Methods:

• Utilized the Canadian Institute for Cybersecurity (CIC) IoT dataset, pre-processed for balanced class representation.
• Developed and compared supervised machine learning models: Random Forest, Adaptive Boosting, Logistic Regression, Perceptron, and Deep Neural Network.
• Applied feature selection, dimensionality reduction, and overfitting minimization techniques to optimize models.

Main Results:

• Random Forest demonstrated optimal performance in both binary and multiclass classification of IoT attacks.
• Achieved an approximate accuracy of 99.55% with the Random Forest model, regardless of feature space reduction.
• Observed a reduction in computational response time, crucial for real-time detection.

Conclusions:

• Machine learning, particularly Random Forest, is highly effective for detecting IoT cyberattacks in healthcare.
• Optimized models provide accurate and rapid identification of network threats, enhancing healthcare IoT security.
• The study highlights the potential of machine learning to safeguard sensitive patient data in connected healthcare systems.