Explainable TabTransformer–RF Model for IoMT Security
Hybrid explainable AI model combining TabTransformer and Random Forest for biometric security in IoMT healthcare systems. Published in Academic Press, 2026.
Overview
The Internet of Medical Things (IoMT) connects medical devices, wearables, and healthcare infrastructure — but introduces significant biometric security vulnerabilities. This project develops a hybrid explainable AI model that combines the representational power of Transformers with the robustness of ensemble methods to secure IoMT systems.
Motivation
IoMT devices collect continuous biometric data (ECG, EEG, fingerprints, etc.). Securing this data against spoofing and unauthorized access requires models that are both accurate and interpretable — clinicians and security auditors must be able to understand why a decision was made.
Methodology
- TabTransformer — applies self-attention to tabular biometric feature data, learning complex inter-feature dependencies
- Random Forest — provides robustness, handles non-linearity, and serves as an interpretable ensemble baseline
- Hybrid architecture — TabTransformer embeddings fed into a Random Forest classifier
- Explainability — SHAP values used to explain individual predictions
Key Contributions
- Hybrid TabTransformer–Random Forest approach for IoMT biometric security
- Explainability layer makes the model audit-ready for clinical environments
- Evaluated on real IoMT biometric datasets with competitive accuracy
Publication
Published in Recent Advances in Computational Intelligence Applications for Biometrics and Biomedical Devices, pp. 285–300, Academic Press, 2026.
Status
Published — Academic Press, 2026.
References
2026
- Academic PressAn explainable hybrid TabTransformer–Random Forest model for biometric security in IoMT Healthcare systemsIn Recent Advances in Computational Intelligence Applications for Biometrics and Biomedical Devices, 2026