Un enfoque de programación dinámica aproximado para la programación de sesiones de quimioterapia en una red de servicios de salud

Abstract

In this work, we study the problem of scheduling chemotherapy sessions in a health system that operates in a network. We consider that patients arrive at one of the treatment centers in the network with the indication of their treating oncologist to carry out one of the different existing treatment protocols. Each treatment protocol uses a certain number of sessions, each with a known duration. In addition, the protocol includes the separation times between the sessions and a maximum waiting period for the first session. The fact that the system works in a network means that the intermediate sessions can be carried out in any of the centers. However, it is required that the first and last session of the protocol be performed at the center to which the oncologist referred the patient. To solve the problem, we model it as a Markovian decision process, which we approximately solve using a linear approximation of the value function. The benefits of the proposed approach are evaluated and compared with the manual approach currently used, simulating a practical case that considers different protocols and two treatment centers in Santiago de Chile. This allows us to assess the benefits of considering a proactive approach (proposal) instead of the reactive approach of the current manual methodology and the impact of coordinated network operation. Our approach allows us to reduce network operating costs by 20% and reduces average waiting times until the first session by approximately half, compared to the manual approach currently used. One of the interesting conclusions of the work is that a treatment system that operates in a network, such as the one proposed, only obtains considerable benefits if it is applied in combination with a proactive policy of scheduling treatment.