The TR&D2 team (Dr. Susan Murphy, Dr. Santosh Kumar, and Dr. Benjamin Marlin) is thrilled to share our latest developments and extend an invitation for collaboration in shaping the future of mHealth interventions. Here’s a glimpse into our recent endeavors:

Tailoring Reinforcement Learning for Dyadic Intervention: This year marks a significant milestone as we worked to develop a Reinforcement Learning (RL) algorithm customized for dyadic interventions.  Our unique setup revolves around a target individual, typically an adolescent or young adult post-bone marrow transplant, and their designated care partner.  With intervention options spanning various time scales and designed to enhance social connection, we’re pioneering the integration of hierarchical RL concepts to address this multifaceted challenge.

Oralytics RL Algorithm Pilot for Dental Health: Dental disease remains a prevalent chronic ailment, prompting the deployment of the Oralytics RL algorithm to improve dental health. Currently in its pilot phase, this initiative aims to provide timely and personalized encouragement for optimal oral self-care behaviors. By leveraging online RL algorithms, we’ve optimized mobile-based prompts to promote effective oral hygiene practices, navigating the complexities of delayed effects and real-world constraints.

Fine-tuning Reward Systems for Optimal Outcomes: A central focus of our efforts lied in designing quality rewards for the RL algorithm. These rewards prioritized desired health outcomes while minimizing user burden.  Through meticulous fine-tuning of reward hyperparameters and simulation environment test beds, we ensured the effectiveness and stability of our RL algorithm within the Oralytics app, an oral self-care app aimed at providing behavioral strategies to bolster patient engagement in oral hygiene practices.

To learn more about the mDOT Center’s TR&D2 and the various research outcomes, visit TR&D2: Optimization.