MedDefender-MHAN: Benchmark Results for Explainable Healthcare IoT Intrusion Detection

Ali Alqazzaz published a dataset on 2026-04-17 containing multi-dimensional comparison results for the MedDefender-MHAN intrusion detection model. The data, stored in an XLS file of 5.5 KB, presents evaluation metrics from tests on CICIDS2017 and TON_IoT benchmark datasets. It includes reported detection accuracies of 99.47% and 98.92%, inference latency, throughput, and explainability alignment scores.

Use Cases

• Benchmarking new intrusion detection models against reported accuracy and latency metrics.
• Analyzing the trade-offs between detection performance and model explainability for regulatory compliance.
• Comparing the effectiveness of embedded explainability mechanisms versus post-hoc methods like SHAP.
• Evaluating model suitability for real-time healthcare IoT security based on throughput and inference speed.

Strengths

• Includes specific, high reported detection accuracies (99.47% and 98.92%) on two benchmark datasets.
• Provides concrete performance metrics like sub-3ms inference latency and 435 samples/second throughput.
• Quantifies model explainability with scores of 94.6% alignment and 91.9% temporal accuracy.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• The dataset is very small (5.5 KB), indicating limited scope, likely containing summary results rather than raw data.

Provenance

Source

Ali Alqazzaz via figshare.

Collection Method

Results from evaluating the MedDefender-MHAN model on benchmark datasets.

Time Range

null

Freshness

Last updated 2026-04-17 17:39:38; freshness should be verified.

Geography

null