This hybrid invited panel will incorporate short talks as well as structured Q&A and focus on highlighting innovative ideas for solving real problems with statistics in health services and outcomes research. Each panelist will give a 10 minute presentation on a cutting edge area of health policy: (1) addressing confounding in machine learning models for imaging data, (2) finding human stories in health care trajectories, (3) mining wearable sensor data for insights into human habits and health, and (4) CancerCLAS: An algorithm and Shiny app for classifying cancer types. This will be followed by a discussion among the panelists with audience Q&A. Planned discussion topics include deploying complex new statistical methods when interpretability is a focus, reproducibility with sensitive data, and how statistics trainees can get started in the health policy domain area. Our panel will provide contemporary insights on the important topic of informing policy with empirical evidence to ultimately improve human health.