A Hybrid Deep Learning and Quantum Optimization Framework for Ransomware Response in Healthcare

This project presents a Hybrid Deep Learning and Quantum Optimization Framework for real-time ransomware detection and response in healthcare environments. The system integrates advanced machine learning techniques with quantum-inspired optimization to protect critical medical infrastructure from PE-based ransomware threats.

The detection pipeline combines Autoencoder for unsupervised anomaly detection and latent feature extraction, Graph Neural Network (GNN) augmentation (with MLP fallback), and TabNet (TabTransformer) for high-accuracy classification on a Windows PE dataset (62,485 samples). Upon detecting a malicious file, the framework performs healthcare-specific impact mapping across hospital assets (ICU, EHR, Radiology, etc.), calculating risk scores based on criticality, patient exposure, and recovery time.

A Quantum Optimization module (QUBO formulation solved via Qiskit or classical greedy solver) generates optimal containment strategies, recommending which systems to isolate under constrained resources. The solution is deployed as a secure Flask web application featuring user authentication, real-time prediction interface, and visualization of containment decisions. This hybrid approach achieves robust ransomware detection while providing actionable, context-aware response planning tailored for healthcare continuity and patient safety.

Keywords: Ransomware Detection, Deep Learning, Autoencoder, TabNet, Graph Neural Network, Quantum Optimization, QUBO, Healthcare Cybersecurity, PE File Analysis, Flask Application, Hospital Asset Prioritization.