| Optimal clinical management of inherited chronic diseases, such as Cystic Fibrosis (CF), requires a dynamic approach which updates treatments to cope with the evolving course of illness and to tailor medicines and dosages for individual patients. The chronic progressive nature of CF and heterogeneity across patients lead to challenges of developing optimal regimens. An adaptive individualized therapy provides a solution and a means toward these goals. In this dissertation, we examine the problem of computing optimal adaptive individualized therapy for CF patients. A temporal difference reinforcement learning method called fitted Q-iteration is utilized to discover the optimal treatment regimen directly from clinical data. We propose multi-state discrete-time Markov process to model the disease dynamic for cystic fibrosis patients with Pseudomonas aeruginosa infection with the model parameters tuned and estimated from the published data in Wisconsin CF neonatal screening project. Our study results indicate that reinforcement learning and the clinical reinforcement trial framework can be an effective tool in discovering and developing personalized therapy which optimises the benefit-risk trade off in multi-stage decision making and improves long term outcomes in chronic diseases. |
