A Comparative Study of AI Search Methods for Personalised Cancer Therapy Synthesis in COPASI

In recent years, optimisation methods in precision medicine have gained much attention thanks to their ability to tackle relevant problems arising from clinical practice effectively. One of the most compelling challenges in this area is designing computational methods for personalising pharmacological treatments, especially for high-impact diseases, due to the large potential impact on the whole healthcare field. In this work, we address the problem of computing safe and effective personalised therapies for Colorectal Cancer (CRC), one of the deadliest forms of tumour for adult humans. We exploit a recent System Biology Markup Language (SBML) mechanistic model of the tumour growth and of the immune response to two drugs and define a simulation-based, non-linear, constrained optimisation problem for automatically synthesising personalised therapies for any given virtual patient. We present a methodology, proposed in our earlier work, that uses a single tool, namely COPASI, to define and solve the optimisation problem. We extend our previous experimental evaluation of the approach by comparing all optimisation algorithms provided by COPASI and performing an in-depth analysis of the results, which provides new and practical insights on the ability of the different algorithms to solve the problem.