Dobi, Balázs: Markov-chain methods in monitoring …

Előadás címe: Markov-chain methods in monitoring health-care processes

Helyszín: Online,

Időpont: 2020.06.29., 14:20 – 14:40

Kivonat: Control charts are widely used tools in industrial and related environments for process monitoring, control and improvement. Health-care processes present challenges which can rarely be found in the traditional industrial settings. The methodology for solving this modelling problem was a Markov-chain based approach, incorporating the costs of illness and medical treatments and the behaviour of the patients. We present the properties of this new, generalized control chart via an application on real-world data. A similar approach can also be applied to chronic disease progression. We created the necessary theoretical background to estimate the parameters governing the propagation between the states and to create a cost-optimal screening strategy as well as creating a simulator for disease progression. This allowed us to check the effect of model misspecification as well as to estimate the estimation errors. We show that under this new framework, estimates for the parameters are stable and accurate.  Moreover, we also show cost-optimal screening policy setup within this framework. The results were presented at several conferences and published in multiple papers and as an R-package (Markovchart).