WLENet: An Workload-Aware Physician Recommendation System with Patient Collective Learning

As a powerful online healthcare service tool, physician recommendation systems enable patients to swiftly and precisely locate suitable physicians at a professional level. However, most existing recommendation systems rarely account for physicians’ workloads, leading to potential inefficiencies in medical resource allocation. To address this practical challenge and research gap, this study proposes a novel deep learning model based on Gated Recurrent Units that explicitly incorporates physicians’ workloads constraints. Specifically, the physician’s workload is calculated by the case complexity of the current consultation and the physician’s remaining consultation capacity. To adapt the GRU framework for this task, a load gate and a time gate are added into to the GRU model, using physician’s workload to control the model’s attention to historical consultation records or the current consultation. Meanwhile, to address the issue of insufficient patient feature information, we develop a patient information enhancement module by extracting collective features from clustered similar patients. The model’s performance is rigorously evaluated using real-world data from Haodf.com, a leading Chinese online health community. Experimental results demonstrate that our model achieves state-of-the-art performance, validating its effectiveness in balancing workload-aware recommendations with accuracy.