RAHM: Relation augmented hierarchical multi-task learning framework for reasonable medication stocking.

•A novel hierarchical multi-task learning framework for reasonable medication stocking.•A relation-aware long short-term memory network (LSTM) to handle the relations of diseases in the progression.•A pseudo residual structure to transmit the useful information along with the hierarchical structure.•Experiments on real-world MIMIC-III dataset demonstrate the advantages of our framework.