Pharmacokinetic--pharmacodynamic blueprints for the anaesthetized adult digital twin: a coupled five-layer specification with an open reference implementation
Year: 2026
Type: article
Abstract: General anaesthesia is a multi-drug intervention, yet the pharmacokinetic–pharmacodynamic (PK/PD) models that guide it are applied one drug at a time. As precision anaesthesia and closed-loop control move toward clinical reality, this fragmentation becomes a central obstacle: no reference maps validated PK/PD models across drug classes onto the clinical signals they drive, specifies the interactions that couple them, and states the result precisely enough to be implemented. This review addresses that gap in three steps. First, the validated single-agent models for hypnotics, opioids, neuromuscular blockers, vasoactive drugs and volatile agents are consolidated into a covariate-aware selection guide that includes the organ-function covariates omitted by anthropometric models. Second, the resulting architecture is converted into a complete mathematical specification: an enumerated state vector with units and initial conditions, an input vector that includes surgical stimulus, the equations of five pharmacodynamic layers (hypnosis, nociception, respiratory safety, haemodynamics, neuromuscular blockade), the mechanistic couplings that link them, and an observation model that treats the bispectral index as a confounded measurement of a latent hypnotic state rather than as that state itself. Third, an open reference implementation supports three in-silico demonstrations: a coupled clinical case, a quantification of what monitor feedback adds to an open-loop population prediction, and a Fisher-information analysis of which parameters are identifiable from which signals. Individualising the twin on twenty minutes of monitoring reduced the median error of a subsequent fifteen-minute-ahead bispectral index prediction from 18.4% to 5.8%, while leaving the underlying effect-site concentration no better estimated, which delimits both the value and the limits of model-based prediction. A prioritised research agenda follows. The framework is a pharmacological substrate and an in-silico test-bench, not a clinically validated device.
Source: Scientific Reviews
Institutions: Technical University of Cluj-Napoca, Ghent University
Language: English
Subfield: Anesthesiology and Pain Medicine
Field: Medicine
Domain: Health Sciences
Open Access status: hybrid
OpenAlex ID: W7225695600