Related Experiment Video
Updated: Jul 25, 2026
08:05
Design and Analysis for Fall Detection System Simplification
Published on: April 6, 2020
Explainable Deep Reinforcement Learning for Anomaly Detection in IoT-Enabled Metaverse Healthcare: Toward Trustworthy
Jing Yang1, Xu Xu2, Muhammad Attique Khan3
1Center of Research for Cyber Security and Network (CSNET), Faculty of Computer Science and Information Technology, Universiti Malaya, Kuala Lumpur 50603, Malaysia.
Research (Washington, D.C.)
|April 27, 2026
Summary
This study introduces a novel cybersecurity framework for metaverse healthcare, utilizing advanced AI for anomaly detection to enhance system security and trustworthiness against cyber threats.
Area of Science:
- Cybersecurity
- Artificial Intelligence
- Healthcare Technology
Background:
- The metaverse offers immersive healthcare experiences but faces significant cybersecurity threats like DoS attacks.
- Existing anomaly detection methods struggle with interpretability, feature selection, class imbalance, and hyperparameter tuning.
- Securing metaverse healthcare analytics requires robust, proactive intrusion detection systems.
Purpose of the Study:
- To develop an advanced anomaly detection framework for Internet of Things-enabled metaverse healthcare environments.
- To address limitations of conventional methods, including poor interpretability and inefficient tuning.
- To enhance the trustworthiness and resilience of healthcare analytics in the metaverse.
Main Methods:
- Implemented an off-policy proximal policy optimization (PPO) algorithm for adaptive, sample-efficient reinforcement learning.
- Integrated SHapley Additive exPlanations (SHAP) for interpretable feature selection.
- Utilized Bayesian Optimization Hyperband for efficient hyperparameter tuning and class-specific reward adjustments for imbalance.
- Evaluated the framework on NSL-KDD, MAWI, and CICIoT2023 datasets.
Main Results:
- The proposed framework achieved state-of-the-art performance across multiple datasets.
- Key metrics demonstrated high accuracy (up to 92.184%), F-measure (up to 88.992%), G-means (up to 89.738%), and AUC (up to 0.873).
- The framework proved effective in identifying cyber threats and showed potential for explainable intelligence.
Conclusions:
- The developed framework effectively enhances cybersecurity in metaverse healthcare applications.
- The integration of PPO, SHAP, and Bayesian Optimization Hyperband addresses key challenges in anomaly detection.
- The results highlight the framework's capability for trustworthy and explainable cyber threat intelligence in dynamic metaverse environments.
