2026 MeTRIC Poster Abstracts

Authors: Jack DiCarlantonio, Jessica Lipschitz, Cara Altimus, Emily Baxi, Veronica Beck, Fernando Goes, Mark Frye, Jennifer Kruse, Angie Lam, Kathryn Lewandowski, Kelly Ryan, Megan Shanahan, Balwinder Singh, Jair Soares, Katherine Burdick, Sarah Sperry

Summary: This study investigates day-to-day variability in sleep and activity (tracked by Fitbit Inspire 3 devices) among individuals with bipolar I disorder, relating this variability to mood fluctuations. A novel data pipeline was created to ensure high quality of Fitbit-derived metrics from 483 participants enrolled so far. Results will guide clinical integration of digital phenotyping in mood disorder care.

Authors: Michael P. Dorsch, Jessica R. Golbus, Rachel Stevens, Brad Trumpower, Tanima Basu, Evan Luff, Kimberly Warden, Michael Giacalone, Sarah Bailey, Gabriella VanAken, Sonali Mishra, Predrag Klasnja, Mark W. Newman, Lesli E. Skolarus, Brahmajee K. Nallamothu

Summary: A 602-participant RCT tested whether a just-in-time adaptive intervention via the myBPmyLife app, supported by Fitbit Versa 2 smartwatches and Omron Evolv BP7000 Bluetooth monitors, could reduce blood pressure. While the intervention improved step count and sodium intake, there was no significant difference in blood pressure reduction versus control—highlighting the need to optimize mHealth use.

Author: Douglas B. Craig

Summary: A bring-your-own-device (BYOD) protocol was developed to capture voice, breath, and cough features (via patient smartphones using their built-in microphones) in asthma care. All audio processing and feature extraction are performed locally on the phone, ensuring privacy. Public datasets are used for benchmarking, and the method aims to provide real-time acoustic biomarkers for clinical decision support.

Authors: Junyoung Park, Neo Kok, Irina Gaynanova

Summary: This study optimizes data-driven thresholds for continuous glucose monitors (CGMs) like Dexcom devices, using population-level data and Python-based analytics. It demonstrates that such thresholds outperform consensus definitions for distinguishing glycemic patterns in people with and without diabetes, and offers an open-source framework for wearable data summarization. 

Code: OptiThresholds GitHub

Authors: Tiffany Sudijono, Irene Tung, Xuhai “Orson” Xu, Emily Capellari, Paloma Vega Martinez, Amanda Kubitz, Arianna Peredia, Marilyn Felix, Elizabeth Guzman, Jennifer Baker, Vanessa Pedroza, Lizbeth “Libby” Benson

Summary: Using surveys and interviews, this mixed-methods study reviewed existing mHealth interventions and co-designed feedback features with expectant parents, particularly those interested in integrations with wearables (e.g., heart rate sensors). Findings show a strong desire for trustworthy and actionable app feedback, especially when augmented with wearable monitoring data to reflect stress and emotion regulation in real life.

Authors: Ziying Wang, Carol Menassa, Vineet Kamat

Summary: Using a simulator, the researchers collected both EEG data (Emotiv EPOC X headset) and wearable sensor data (Shimmer3 GSR+ for skin conductance, optical PPG for heart rate), teaching a model to use only wearable/mobile signals to predict driver readiness for automated driving takeover, eliminating the burden of EEG collection in real-world scenarios.

Authors: Allison Doroshewitz, Rachel Garabedian, Kira Birditt, Noelle Carlozzi, Sung Choi

Summary: In this pilot with dementia caregivers, participants used the Roadmap 2.0 mHealth app and wore Fitbit Inspire devices to log positive activity, sleep, and physical activity over two weeks. Engagement with app activities was associated with better mood and lower depression and anxiety. The app was well received, paving the way for larger trials.

Authors: Madeline Abbott, Rachel Gonzalez, Walter Dempsey, Michael Dorsch, Scott Hummel, Brahmajee Nallamothu

Summary: Researchers enhanced a reinforcement learning algorithm for just-in-time adaptive interventions, as tested in the LowSalt4Life (LS4L) mobile app that sends push notifications (geo-fenced by user smartphones) to reduce sodium intake. The approach allows treatment adaptation based on user context and copes with binary behavioral outcomes typical in mHealth MRTs.

Authors: Owen Yoo, Irina Gaynanova

Summary: “ihr” is an open-source R package designed to analyze heart rate variability data from wearable devices such as Fitbits and Apple Watches, leveraging methods from continuous glucose monitoring to provide detailed and reproducible cardiovascular patterns beneficial for health research and clinical applications. 

Code: ihr GitHub

Authors: Neo Kok, Walter Williamson, Joyce Lee, Irina Gaynanova

Summary: Analysis of 1,010 Dexcom CGM profiles revealed that standard 14-day monitoring (versus 90 days) and data loss both degrade precision and introduce bias in diabetes research. The authors recommend longer monitoring (minimum 30 days) or sample size adjustments when using CGM as the primary wearable technology in clinical studies.

Authors: Juniar Lucien, Krista Noviski, Danielle Czarnecki, Sam Chuisano, Melissa DeJonckheere

Summary: This study utilized Dexcom G6 and G7 CGM devices, along with narrative text message surveys (delivered to participants’ phones), to gather detailed, real-time data on how 40 adolescents managed their diabetes daily. Integration of wearable-derived glucose streams and mobile-based qualitative self-reports provided a holistic view of technology use and daily decision-making.

Authors: Valerie Mefford, Alexandra Geiger, Ibrahim Khan, Maryann Riggs, Paul Schenk, Michele Boertman, Matt Konerman, Monika Leja, Jeff Kullgren

Summary: A remote patient monitoring (RPM) program for hypertension offered validated Bluetooth-enabled BP cuffs (linked via patients’ mobile phones) and virtual care to reduce cardiology clinic waits. This approach provided faster clinical contact and high rates of BP control in the RPM group compared to standard care.

Authors: Kai He, Yu Fang, Amy Bohnert, Srijan Sen, Meng Wang

Summary: Medical interns wore Fitbit smartwatches (various models, including Charge 2 and Charge 4) to passively track sleep and activity, with personalized mood correlations derived from these metrics. An LLM-powered platform (MoodDriver) delivered data-driven, individualized feedback via mobile app messages to support mood improvement based on objectively tracked behavior.

Authors: Kaiyuan Wang, Dorcas Mwigereri, Andrew Aballa, Amos Bunde, Eileen Weinheimer-Haus, Jessica Baker, Akbar Waljee, Elena Frank, Zhenke Wu

Summary: Over 12 months, 514 Kenyan healthcare workers wore Fitbit 2 wearables and submitted daily mood via a mobile app. Machine learning predictions showed self-rated mood was the strongest predictor of future depression, while wearable features (sleep/steps/heart rate) were also modestly helpful, supporting the use of wearables for proactive mental health monitoring.

Authors: Gjulia Camaj, Victoria Murphy, Julia L. Smith, Sarah Sperry

Summary: Across 107 participants (bipolar disorder, subclinical, and healthy), sleep assessed objectively with Fitbit Charge 5/6 devices (data extracted via Fitabase) was analyzed alongside neuropsychological testing. Preliminary analysis explored which aspects of wearable-measured sleep disturbance best predict cognitive function, informing clinical intervention points.

Authors: Leyuan Qian, R. Nisha Aurora, Naresh M. Punjabi, Irina Gaynanova

Summary: Using data collected by Welch Allyn ABPM 6100 BP monitors and Philips Actiwatch actigraphy devices, the “SmoothHOOI” R package was developed for temporally-aware tensor decomposition of ambulatory blood pressure and heart rate. The method uncovered finer temporal structures associated with patient characteristics (e.g., sleep apnea severity) compared to summary statistics. 

Code: SmoothHOOI GitHub

Authors: Aishani Kulshreshtha, Yu Fang, Elizabeth D. Mills, Amy S.B. Bohnert, Srijan Sen

Summary: This study tracked 1,476 mental health patients’ sleep and physical activity using Fitbit wearables, alongside daily mood logged in the MyDataHelps app. Analysis found individualized, bidirectional links between sleep duration, physical activity, and mood, supporting the advantage of mobile and wearable monitoring for personalized mental health support.

Authors: Keara Sullivan, Abigail Bretzin, Doug Wiebe, Nichole Burnside

Summary: Through clinician focus groups, this project explored concussion management in NCAA sports, including how practices could improve with real-time app-based platforms (like Recoups/Headcheck), which streamline symptom entry and tracking via mobile/EHR technologies. Current barriers identified included clinician time, athlete volume, and resource limits.

Authors: Victoria Murphy, Gabi Skinner, Irmgard Pallas, Steve Andreau, Melvin McInnis, Sarah Sperry

Summary: Forty-seven bipolar spectrum disorder participants completed frequent EMA surveys (delivered via the MyDataHelps mobile platform) to capture time spent alone, substance use, and affect instability. Models revealed cannabis use, time alone, and mood instability predicted clinically assessed suicidal ideation, showing the potential for smartphone-based EMA to track suicide risk in real time.

Authors: Andrew Aballa, Dorcas Mwigereri, Zhuo Zhao, Willie Njoroge, Linda Khakali, Rachel Maina, David Andai, James Orwa, Amos Bunde, Eileen Weinheimer-Haus, Jessica Baker, Lukoye Atwoli, Srijan Sen, Akbar Waljee, Anthony Ngugi, Amina Abubakar, Zhenke Wu, Zul Merali, Elena Frank

Summary: In a year-long longitudinal study, 514 Kenyan HCWs used the MyDataHelps app (for quarterly surveys) and Fitbit Inspire wearables. Over 43% met major depression criteria at some point during the year, with risk strongly related to stress, discrimination, and work experience, highlighting a need for culturally appropriate mobile interventions.

Authors: Cole Geschwender, HoWon Kim, Amanda Melvin, Ethan Lajiness, Zheng-Yang Zhao, Adam Lepley

Summary: Using Garmin Forerunner 55 wearables, this study found that people with prior ACL reconstruction took significantly fewer steps, did less vigorous activity, and had lower daily energy expenditure than controls, despite similar cardiorespiratory fitness and knee bone mineral density (measured via iDXA). Results guide rehabilitation care and monitoring strategies.