Ensuring timely and consistent access to medical test results is crucial for improving the efficiency of healthcare providers. For example, doctors need to know when test results will be available to plan potential follow-up care, which has knock-on effects throughout the healthcare system. However, the variation in the types of tests, the fluctuations in demand for different tests, and the complexity of the associated logistical processes make providing such speed and consistency challenging. In this project, we will analyse a variety of real-world inspired problems that need to be overcome to meet this challenge together with Labmicta, a leading medical laboratory in the Twente region in the Netherlands.
As an example, consider the problem of planning the (multi-step) processing of a medical test under uncertainty regarding the demand that will have to be met in the near future. This requires building robust models to predict future demand, making decisions under time-heterogeneous uncertainty, and adapting to the state of the processing system online. Our approach will be to draw on a broad selection of tools, including (deep) reinforcement learning, queuing networks, online algorithms, and systems engineering. In addition, a large component of the project is to ensure that the results translate into impact by ensuring that they are implemented in the real-world in the Labmicta laboratory management systems.