A Patient-Specific, Dynamic, Multiobjective Model for Hepatocellular Carcinoma Treatments

This study addresses a systems-based approach for healthcare delivery to patients with hepatocellular carcinoma (HCC), the most common type of liver cancer, who are awaiting orthotopic liver transplantation (OLT). Often waiting as long as a year for OLT, these patients undergo many intermediary treatments, the sequence of which uniquely changes over time for each patient. These combinations of therapies are too many in number for a physician to evaluate accurately in his mind, as he must administer the most effective therapy at each checkup. To overcome these limitations, this study describes a dynamic multiobjective decision tree (D-MODT) designed by the author to provide medical decisionmakers with Pareto-optimal treatment strategies for patients suffering from HCC. A multistage decision tree framework is developed, allowing the model to be customized for each patient in terms of state of health and geographic location by updating via Bayes’ Theorem a patient’s probabilities of survival, progression, etc. as every piece of new information arrives to the physician. This level of refinement has not achieved, to the author’s knowledge, in any study employing the traditional Markovian approach, where only transitions from the most recent state of health are considered. The model’s framework and mathematics are described herein, along with a case study that details how a 1-month case of the model would be solved and analyzed. The model is populated with data from UNOS and the relevant literature, although data collection and better population of the model are large future directions for this research. Although this methodological approach is herein applied to HCC treatment, its success should encourage its application in other areas of medicine where complex sequential decisionmaking confounds physicians in disease management scenarios.