RASnet: Recurrent Aggregation Neural Network for Safe and Efficient Drug Recommendation

Drug recommendation is one of the most important research topics in smart healthcare. Its goal is to provide a set of safe drug combinations based on the patient’s electrical medical records(EHR). Drug recommendation is challenging since it is difficult to obtain an appropriate representation of patients' health states from these personalized historical records. Meanwhile, drug recommendation must prioritize the safety of drug combinations because drug-drug interactions(DDI) can result in side effects. In order to address these issues, we propose a novel recurrent aggregation neural network for safe drug recommendation, called RASNet. RASNet introduces a straightforward but efficient recurrent aggregation neural network to capture historical records related to the patient’s health state of current visit, which can improve the performance of EHR-based healthy state modeling, particularly in cases when the patient’s condition changes periodically. Furthermore, this paper presents a novel exponential controller for DDI to enhance the safety of drug combinations. Our proposed DDI controller not only balances the DDI rate between the safety and accuracy of drug recommendation but also ensures performance even when the DDI rate is low. Extensive experiments on the MIMIC-III dataset demonstrate that RASNet achieves state-of-the-art performance. Moreover, RASNet exhibits excellent efficiency and safety in drug recommendation.