Do Outpatient Services Impact Length of Stay? A Preliminary Analysis through Deep Learning Classification

Timely discharge of hospitalized patients can prevent complications and reduce costs. In this paper, we have investigated whether outpatient services, i.e., diagnostic exams or specialist consultations provided by external wards, have a clear impact on Length of Stay (𝐿𝑂𝑆). In particular, we have worked on an event log of more than 7000 real patient traces, logging the sequence of outpatient services provided during hospitalization, and we have classified the traces into long (≥ 20 days) versus short (< 20 days) 𝐿𝑂𝑆, resorting to a deep learning approach that adopts a Long Short-Term Memory (LSTM) network. The very high quality of the classification results suggests that outpatient services play a significant role in determining 𝐿𝑂𝑆, and hospitals should work on optimizing their organization.