To develop and validate, a real-time algorithm to individualize unfractionated heparin (UFH) dosing based on activated clotting time (ACT) dynamics in cardiac procedures with or without cardiopulmonary bypass (CPB).

Retrospective cohort study

University hospitals; multi-institutional for external validation.

Adult patients undergoing cardiac surgery with or without CPB.

Development of a pharmacodynamic model of ACT response to UFH incorporating CPB-adjusted scaling. Based on this model, an individualized dosing algorithm was implemented to enable a real-time online decision-support tool for clinical use.

The model was developed using data from 446 patients (2534 ACT measurements) and externally validated in 105 patients from an independent center. Clinical implementation was retrospectively evaluated in 83 patients and compared with 83 matched historical controls. The model demonstrated strong predictive performance, with decreasing bias and mean absolute error as additional ACT measurements were incorporated. Individualized dosing achieved ACT values closer to target and significantly reduced ACT variability compared with controls (variance ratio = 2.3; p < 0.001), while using lower UFH doses without compromising anticoagulation.

The real-time individualized UFH dosing algorithm was associated with improved control of ACT values and reduced variability during cardiac procedures in this retrospective analysis. These findings suggest that Bayesian-guided dosing may enhance the precision of anticoagulation management under routine clinical conditions. However, given the study design, device-specific ACT measurements, and limited clinical outcome data, these results should be interpreted as preliminary. Prospective, adequately powered studies are required to confirm clinical benefit, safety, and broader applicability.