Publication Details
Abstract
We propose a nationwide AI-Driven, HIPAA-Aligned Cyber-Clinical Business Intelligence System (CCBIS) that will improve disease-risk prediction and prognostication, bolster cyber-defenses in healthcare, render more efficient IT and OPEX (operational expenditure) investments at U.S. healthcare firms across the continuum of care. Our framework combines stochastic differential equations to describe individual patient-level health dynamics, graph-based networks of patients for modeling population relationships and machine-learning in order to enhance predictive accuracy. A multi-objective optimization model is further formulated to optimize both predictability performance, cybersecurity robustness and budget efficiency in a real-life budget constraint scenario. Results Experiment results on large-scale synthetic datasets (which are calibrated, the ground truth US healthcare settings) indicate significantly improved prediction accuracy in disease risk, as well as early detection of cyber-attacks, reduction and escalation ate in breach rates and more targeted allocation of IT/operational resources. The results show that the CCBIS is an efficient, scalable, mathematically sound and policy-driven approach for ensuring a secure, predictive and cost-effective health decision-making at the national level.