Personalized Decentralized FOPI Control for Hypnosis-Analgesia Regulation in Anesthesia

This paper presents a personalized decentralized fractional-order PI (FOPI) control framework for a 2 × 2 multipleinput multiple-output (MIMO) anesthesia system regulating Bispectral Index (BIS) and Nociception Level (NOL) via Propofol and Remifentanil infusion. The approach relies on patientspecific pharmacokinetic-pharmacodynamic modeling using response surface models (RSMs), enabling analytical computation of unique steady-state operating points while enforcing common clinical targets. RSMs are locally linearized around these patientspecific equilibria, yielding individualized linear models that form the basis for controller synthesis. Decentralized FOPI controllers are subsequently designed using a frequency-domain autotuning methodology that enforces gain crossover frequency, phase margin, and iso-damping specifications. The proposed personalized framework is evaluated in simulation on a cohort of ten virtual patients exhibiting significant variability in pharmacodynamic response. Simulation results demonstrate consistent reference tracking, reduced inter-patient variability during induction, and robust disturbance rejection during maintenance.