Towards Optimizing Hospital Patient Transports by Automatically Identifying Interpretable Causes of Delays

Abstract

The continuous financial pressure on hospitals forces them to rethink various workflows. We focus on optimizing hospital transports, within the hospital, as they count up to 30% of the overall hospital cost. In this paper, we discuss a self-learning platform that learns the causes of transport delays, in order to avoid these kinds of delays in the future. We pay special attention to the explainability of the self-learning system, such that management understands the learned causes and remains in control over the automated process. This is achieved by providing the learned causes as sentences that can be understood by non-technical personnel and allowing these causes to first be supervised before the system takes them into account. Once approved, the system will calculate how much more time should be assigned to these transports in order to avoid future delays. As a result, the scheduling of patient transportation can be automatically optimized, while management remains in full control of the process.