What is a Markov Model and why are they used for health economics?
Markov models allow you to model disease progression over time. For example, you might be modeling cancer which develops and progresses slowly over years. If you tried to model this with a simple Decision Tree rather than a Markov Decision Tree, you would have hundreds or thousands of nodes within the model to account for progression and other events in each year or month.
Markov models handle this seamlessly with a framework of health states and events. Patients start each cycle in a health state, then pass through event patient pathways within a time cycle. The patients are then sent back to the same or different states to start the next cycle. This allows a single progression event in the model to occur for different patients at different times in the future.
While the patients pass through the health states and events, they accumulate costs and utilities. Depending on the state, the cost for chronic care and the utility will be different. Costs may also be accumulated for events like hospitalization or screening. Markov Cohort Analysis accounts for the percentage of the overall patient cohort that passes through these health states and events to apply a weighted percentage to those cost/utility entries. The resulting sum of all weighted cost/utility values for all cycles is the average cost/utility per patient.
Markov models for health economics are frequently used to compare treatment/screening/diagnosis strategies for a specific health problem. These health economic models are run to calculate the average cost/utility of all strategies, which are then compared via cost-effectiveness analysis. This determines the most cost-effective strategy for the specific health problem.