Capacity Allocation Of Critical Resources In The Hospital

Imaging facilities are usually critical resources in the hospital.With the increase of demand for imaging inspection,Managers are under great pressure to make maximum use of these resources,which leaves patients on long waiting lists,possibly increasing their anxiety and aggravating their conditions.The fact that some patients do not show up for their appointments makes this problem more challenging.This paper studies the dynamic system of capacity allocation by considering both different patients’ waiting time targets(WTT)and no-shows.Patients with different priorities randomly arrive over time and decisions are made regarding the number of different priority patients scheduled on which days within patients’ type-specific WTTs.This paper firstly study a simple markov decision process model which only consider two types of patient and then extend to multi-priority patient’s capacity allocation problem,and finally,include no-show into consideration.This paper introduces bi-dimensional nested protection level policies,which are nested in the sense that more capacity is reserved for higher-priority patients on a same day and less capacity is reserved for the same priority patients on days earlier in the booking horizon.The criterion function is to minimize the weighted number of patients who can not be served within type-specific WTT and overtime penalty.By using partial derivatives of the criterion function with respect to the protection level as the descending direction,this paper uses sample-path to estimate the partial derivatives and proposes a stochastic approximation-based algorithm that combines steepest descent and simulated annealing to improve the protection levels.Numerical results show that the proposed solution algorithm can greatly reduce the average daily referral and overtime penalty than simulated annealing and greedy heuristic.The proposed algorithm produces slightly smaller average daily referral and overtime penalties than the steepest descent method,with far less noise.