Implementing An Integrated Self-Decision Making and Self- Optimization Framework for Autonomic Medical Cyber-Physical Systems

Medical Cyber-Physical Systems (MCPS) integrate sensing, computation, and actuation for healthcare tasks (e.g. patient monitoring, diagnostics). Their complexity and safety-critical nature demand autonomous decision-making under uncertainty and real-time self-optimization. We propose a unified framework combining edge computing, AI-driven decision engines, multi-objective optimization, and digital twins to achieve fully self-managing MCPS. The system collects real-time patient data via wearable sensors, preprocesses them at the edge, and employs a decision engine (reinforcement learning, fuzzy logic, multi-criteria decision analysis) to determine control actions (e.g. alert generation, therapy adjustments). A digital-twin of the patient and environment simulates scenarios to refine decisions and optimize long-term objectives (accuracy, latency, energy). We present the mathematical formulation (MDP for decision, weighted-sum multi-objective optimization), detailed algorithm and an evaluation using arrhythmia detection as a case study. Experimental results on ECG datasets demonstrate that our hybrid framework yields superior detection accuracy (e.g. 95.1% sensitivity, AUC 0.96) and real-time performance (decision latency approx 42 ms) compared to baseline methods. This integrated approach advances the autonomics of MCPS by achieving statistically significant improvements in robustness, adaptability, and efficiency