A model for optimal sequential decisions applied to liver transplantation

The present investigation aims at the construction of a sequential decision rule derived from a decision-theoretic model. An optimal clinical management strategy based on patient risk assessment is to be found. A novel decision theoretic cost model has been selected for this clinical classification task: costs were determined by specifying a minimum acceptable sensitivity and specificity of the overall procedure. Non-linear optimisation combined with a robust partial classification method is used to find the earliest possible decision step where a final decision can be made subject to these quality restrictions. The probabilities needed in the model are estimated from data provided by a clinical study on liver transplantation patients. Decision steps for the decision-theoretic model were chosen before and after donor organ assessment, and postoperatively in the intensive care unit. Clinical parameters acquired in between decision steps were combined to scores obtained from artificial neural networks (ANNs). The encouraging results show the applicability of the model in the clinical setting.