Susan Murphy, Mallinckrodt Professor of Statistics and of Computer Science and associate faculty at Harvard’s Kempner Institute for the Study of Natural and Artificial Intelligence, will present the 2026 John M. Chemerda Lectures in Science at Penn State University Park on 4 and 5 November.
Two lectures aimed at different audiences
The series includes a public lecture and a more technical scientific talk. Both events are free and open to the public, and are sponsored by the Penn State Eberly College of Science.
- Public lecture: “Online Reinforcement Learning in Personalizing Digital Health Interventions” — 4 November, 16:00, 114 McAllister Building.
- Scientific lecture: “Replicability in Bandit and Reinforcement Learning Algorithms” — 5 November, 15:30, 201 Thomas Building.
The public lecture will outline challenges and emerging solutions for using online reinforcement learning (RL) in digital health systems. According to the lecture description, these systems aim to personalise support for patients dealing with conditions such as substance misuse, hypertension and outcomes after bone marrow transplantation. Key difficulties include learning when and where to offer support while the individual is receiving interventions and evaluating the RL algorithm’s usefulness across repeated deployments.
Research background and methods
Murphy’s research concentrates on improving sequential, individualised decision-making in digital health. She is credited with developing the microrandomised trial, a trial design used to construct and assess just-in-time adaptive interventions delivered through mobile and digital platforms. Her laboratory focuses on developing and deploying online learning algorithms that adapt to individual users in real time.
| Lecture | Date | Time | Venue |
|---|---|---|---|
| Online Reinforcement Learning in Personalizing Digital Health Interventions | 4 November 2026 | 16:00 | 114 McAllister Building |
| Replicability in Bandit and Reinforcement Learning Algorithms | 5 November 2026 | 15:30 | 201 Thomas Building |
Microrandomised trials randomise individuals repeatedly over time to different intervention components, enabling researchers to estimate the immediate effect of a treatment decision and to gather data for algorithms that adapt to changing circumstances. These designs have been adopted across a range of health-related digital interventions, the lecture notes state.
Why the lectures matter
The talks bring attention to two interlinked technical problems in digital health: how to personalise interventions effectively using online RL, and how to ensure results are replicable when algorithms and deployments vary. Replicability — the capacity to reproduce algorithmic performance across settings and studies — is a growing concern across machine learning applications in healthcare, where deployment contexts, user behaviour and data collection protocols can differ substantially.
For health systems considering mobile and digital interventions, the methodological advances Murphy discusses are relevant to design choices, evaluation strategies and regulatory scrutiny. Microrandomised trials and online RL methods aim to produce evidence that is both actionable for individual patients and robust enough to inform broader policy and clinical practice.
All members of the public are invited to attend either lecture, offering an opportunity for clinicians, researchers, policymakers and technologists to engage directly with current methodological work at the intersection of statistics, artificial intelligence and digital health.
Event organiser: Penn State Eberly College of Science.
