Semi-empirical Anticoagulation Model (SAM): Dose prediction during warfarin therapy.

Management of oral anticoagulant therapy is essential to ensure a good quality of life for patients. To assist clinicians, several computerized dosing algorithms were developed to determine optimal anticoagulant doses. However, these algorithms have several limitations that can lead to inaccuracies in dosing recommendations. To overcome some of these challenges, this paper proposes a warfarin dose prediction algorithm using individual patient sensitivity analysis. The proposed algorithm is designed to account for factors that influence an individual's response to warfarin, allowing it to more accurately adjust and optimize dosing. It is based on the Semi-empirical Anticoagulation Model (SAM), but has been extended to incorporate stochastic process analysis. This allows the algorithm to assess the probability of events that could change the relationship between the International Normalized Ratio (INR) and the therapy administered. To evaluate the effectiveness of the algorithm, a retrospective observational study of 1796 patients was conducted over one year. The study compared the doses of warfarin administered in clinical practice with those suggested by the algorithm. Percentage Bland-Altman Analysis was used to assess accuracy, which showed that the algorithm had an average accuracy of (3.24±25.80)%. When compared to other algorithms in the literature, which showed an accuracy of (5.73±60.9)%, the proposed algorithm showed significantly better accuracy. The improved accuracy of the proposed algorithm allows for more flexible and precise adjustments to therapy, resulting in an INR closer to the target range with less variability. This ultimately improves patient safety and the overall quality of life for those suffering from venous thromboembolism.