Explore a 30-minute oral presentation from the Uncertainty in Artificial Intelligence conference that delves into the challenges of inferring causal effects of continuous-valued treatments from observational data. Learn about a novel methodology for bounding average and conditional average continuous-valued treatment-effect estimates when hidden confounding prevents point identification. Discover how this approach provides tighter coverage of true dose-response curves compared to existing models and baselines, as demonstrated through semi-synthetic benchmarks on multiple datasets. Gain insights into the application of this method in a real-world observational case study, highlighting the importance of identifying dose-dependent causal effects in decision-making processes.