How to Predict Relapse in Leukaemia Using Time Series Data: A Comparative in silico Study

Risk stratification and treatment decisions for leukaemia patients are regularly based on clinical markers determined at diagnosis, while measurements on system dynamics are often neglected. However, there is increasing evidence that linking quantitative time-course information to disease outcomes can improving the predictions for patient-specific treatment response.We analyzed the potential of different computational methods to accurately predict relapse for chronic and acute myeloid leukaemia, particularly focusing on the influence of data quality and quantity. Technically, we used clinical reference data to generate in-silico patients with varying levels of data quality. Based hereon, we compared the performance of mechanistic models, generalized linear models, and neural networks with respect to their accuracy for relapse prediction. We found that data quality has a higher impact on prediction accuracy than the specific choice of the method. We further show that adapted treatment and measurement schemes can considerably improve prediction accuracy. Our proof-of-principle study highlights how computational methods and optimized data acquisition strategies can improve risk assessment and treatment of leukaemia patients.