Year 1 (2019 - 2020) Projects
Reinforcement learning to personalize message framing for health habits (REINFORCE)
This pilot seeks to evaluate the feasibility and effectiveness of a machine-learning based approach to optimize and personalize medication adherence health communication for patients with diabetes. The project consists of a pragmatic trial. We also seek to cluster intervention
patients by their response to different types of messages and evaluate the ability to predict these cluster phenotypes using information prior to randomization.
Using Cues and Rewards in Patients with Arthritis and Rheumatic Disease
This pilot aims to evaluate the feasibility and effectiveness of a habit formation intervention to motivate medication adherence for patients with rheumatic conditions. We are conducting a pragmatic randomized trial to determine whether strategies that couple context cues and
intermittent rewards improve medication adherence in rheumatic patients.
Co-PIs: Julie Lauffenburger, PharmD, PhD, and Punam Keller, MBA, PhD
Co-PIs: Candace Feldman, MD, MPH, ScD, and Wendy Wood, PhD
Year 2 (2020 - 2021) Projects
Personalizing intervention to reduce clinical inertia in the treatment of hypertension
This pilot seeks to empirically identify factors contributing to clinical inertia and to identify the relationship between these factors and intervention responsiveness. To achieve this, we will test two interventions targeting inertia (social norming and academic e-detailing) compared to control in a pragmatic, randomized pilot study. We will then identify intervention responsiveness profiles through patient and physician clustering using a machine learning approach.
Novel application of simulation for providers to overcome the hot-cold empathy gap in high-risk medication prescribing
This pilot aims to reduce the prescribing of high-risk medications, such as antipsychotics and benzodiazepines, to hospitalized older adults. We hypothesize that the use of these medications is driven by "system 1" thinking and that prescribing can be improved using simulation training. We are conducting a pilot randomized trial to evaluate the effectiveness of the simulation approach that we developed.
Co-PIs: Nancy Haff, MD, MPH, and Niteesh Choudhry, MD, PhD
Co-PIs: Julie Lauffenburger, PharmD, PhD, and Niteesh Choudhry, MD, PhD
Year 3 (2021 - 2022) Projects
Use of construal level theory to inform messaging to increase vaccination against COVID-19
This randomized trial tests the effects of "how" versus "why: message framing on COVID-19 vaccination rates using electronically delivered communication, followed by analyses to identify patient characteristics that might predict intervention responsiveness to allow for further tailored communication after the completion of the trial.
Leveraging social networks to control hypertension after ischemic stroke
This project aims to leverage social networks to improve hypertension control. We are conducting a randomized trial to determine the effects of a social networks intervention versus individual hypertension counseling on blood pressure control for patients discharged after ischemic stroke. We also aim to determine the shortest questionnaire that maximizes network information important for the intervention versus the previously validated full version.
Co-PIs: Nancy Haff, MD, MPH, and Julie Lauffenburger, PharmD, PhD
PI: Amar Dhand, MD, DPhil
Breaking implicit bias habits: An individuation pilot to promote equity in rheumatic disease care
PI: Candace Feldman, MD, MPH, ScD
The goal of this pilot is to test the efficacy of an individuation-based intervention among rheumatologists and racially discordant patients at two large, multisite practices to improve racial and SES equity in receipt of high-quality care. We will conduct a pragmatic randomized trial to test the efficact of a real-time, provider-based individuation intervention versus standard of care to improve the receipt of high-quality rheumatic disease care among Black and lower socioeconomic status individuals. We will also evaluate the effect of the intervention on provider-patient communication, adherence, provider trust and care satisfaction.
Year 4 (2022 - 2023) Projects
Refinement and adaption of reinforcement learning to personalize behavioral messaging for healthy habits
This project aims to expand upon the previously funded REINFORCE year 1 Roybal project. We will refine and adapt a novel reinforcement learning-based text messaging intervention within a community setting (NIH Stage I activity). We then will conduct a small, real-world efficacy randomized pragmatic trial to determine whether the adapted reinforcement learning-based intervention improves medication adherence within the community setting (NIH Stage III activity).
Addressing diffusion of responsibility and prescribing burden to improve use of diabetes medications
This project aims to address the underuse of sodium-glucose cotransporter-2 inhibitor (SLGT-2is) and glucagon-like peptide-1 receptor agonist (GLP-1RAs) medications for diabetes in primary care. We will conduct a randomized trial among primary care providers to test the impact of an intervention targeting diffusion of responsibility with and without additional resources to address reduction of prescribing burden. We will then Identify characteristics of the patient, provider, and their clinical interaction that are associated with responsiveness to each intervention.