Introducing scan budgets to optimize allocation of radiology capacity

We study a scan budget system as a solution to the problem of overload for a radiology department. To be able to deal with fluctuations of the demand, we introduce the possibility of trading: mutual exchange of budget among departments. The decentralized multi-agent problem is modeled using single-agent Markov Decision Processes that are connected via trading probabilities. We analyze different individual policies under a penalty and reward mechanism for trading and analyze the resulting collective behaviour using acceptance ratios. We find that the possibility of trading increases the acceptance ratios for all scan priorities and reduces the waste of capacity and the average annual departmental difference. When the number of trading possibilities is large, the acceptance ratios approach the acceptance ratios under centralized admission control.