Dynamic Causal Modeling of Heart Rate–Blood Pressure Coupling During Induction of Anesthesia: A Systems-Based Approach to Predict Vasopressor Requirements

Background: Induction of general anesthesia frequently induces hemodynamic instability characterized by hypotension, necessitating timely administration of vasopressors. The causal relationship between heart rate (HR) and blood pressure (BP) during this critical period remains incompletely understood, limiting the development of predictive models for vasopressor requirements. Objective: This study employed dynamic causal modeling (DCM) to characterize the directional interactions between HR and BP during anesthetic induction and to develop a systems-based predictive framework for vasopressor needs. Methods: We prospectively enrolled ۱۲۰ adult patients undergoing elective major surgery under general anesthesia. Continuous HR and invasive arterial BP data were recorded from ۵ minutes pre-induction to ۱۰ minutes post-intubation. A time-varying Granger causality analysis was applied to quantify the feedforward (HR→BP) and feedback (BP→HR) causal pathways. A dynamic Bayesian network integrating hemodynamic, pharmacodynamic, and patient-specific parameters was constructed to predict vasopressor requirements. Results: Propofol induction significantly attenuated the feedback pathway (BP→HR) from baseline (causal coefficient: ۰.۴۲±۰.۱۱ vs. ۰.۱۸±۰.۰۹, p<۰.۰۰۱) while preserving the feedforward pathway (HR→BP). The magnitude of feedback attenuation correlated strongly with subsequent vasopressor dose (r=۰.۷۳, p<۰.۰۰۱). The DCM-based prediction model achieved an AUC of ۰.۸۹ (۹۵% CI: ۰.۸۳-۰.۹۴) for predicting vasopressor requirements >۵۰ µg phenylephrine equivalent.Conclusion: Dynamic causal modeling of HR-BP coupling during anesthetic induction provides mechanistic insights into hemodynamic instability and enables accurate, individualized prediction of vasopressor requirements, offering a promising framework for precision hemodynamic management.

کلیدواژه ها:

dynamic causal modeling ، Granger causality ، anesthesia induction ، vasopressor prediction ، hemodynamic monitoring