Deep Reinforcement Learning Control of Type-1 Diabetes with Cross-Patient Generalization | IEEE Conference Publication | IEEE Xplore