Előadás címe: Multivariate Markov-chain methods in monitoring health-care processes
Időpont: 2020.06.30., 10:35 – 10:55
Kivonat: The application of Markov chain-based cost-optimal control charts on real-word data revealed that the method can be extended in areas where information is naturally multivariate (e.g. treatments, demographic variables etc.). From a mathematical perspective, it means that the so far three-dimensional parameter space needs to be expanded into higher dimensions and that it may include discrete/categorical variables. We have shown an application example which highlights the importance of multidimensionality. Our main goal to pursue in the disease progression setup is related to the above one. Incorporating more (possibly multivariate) information should allow for more exact estimation of the cost-optimal policy of screening. The second goal is to study the identifiability of parameters in disease progression models, we aim to establish the theoretical reasons behind the flaws of convolution based models. These are constructed by assuming the sojourn time and the preclinical intensity to be random variables with parameters to be estimated. However, initial results show that these parameters are not always identifiable. The extended models are planned to be used in epidemic monitoring scenarios (e.g. flu, COVID-19) too with the Healthware Consulting Ltd.