Poster Session Abstracts

Authors: Jiuchen Zhang, Fei Xue, Qi Xu, Jung-Ah Lee, Annie Qu

Background: Mobile health has emerged as a major success for tracking individual health status due to the popularity and power of smartphones and wearable devices. However, this has also brought great challenges in handling heterogeneous, multi-resolution data, which arise ubiquitously in mobile health due to irregular multivariate measurements collected from individuals. In this paper, we propose an individualized, dynamic latent factor model for irregular multi-resolution time series data to interpolate unsampled measurements of time series with low resolution.
 

Authors: Lan Shi, Marianthie Wank, Yan Chen, Yibo Wang, Yachuan Liu, Emily C. Hector, and Peter X.K. Song

Background: Wearable health devices have drawn wide attention as healthcare shifts towards personalized medicine and individuals seek to monitor their health. These devices enable real-time health monitoring, which can provide new information for determining and predicting health conditions and preliminary medical diagnoses. Understanding the different ways in which we can analyze the vast amounts of data collected from these devices is of increasing importance. In this project, we focus on a classification analysis of sleep status using physiological data collected by the E4 Empatica wristband.
 

Authors: Michelle Segar, Kathy Poore, Reema Kadri, Molly MacDonald, Rob Adwere-Boamah, Juno Orr, Lorraine Buis and Zora Djuric 

Background To change behavior in ways likely to be sustained, we need to go beyond educating people about facts and stats to influence their mindsets about behavior. A mindset constitutes a set of beliefs, attitudes, and perspectives that shape people’s thinking. An individual’s mindset reflects their fundamental orientation toward any behavior. Mounting research shows that mindset plays a powerful role in psychological, behavioral, and physiological outcomes, and that interventions can change mindsets to be more adaptive. To change the mindset toward healthy eating in ways that might support long-term change, interventions should aim to create key mental shifts related to motivation, self-regulation, and learning.

Authors: Umaiyal Kogulan, Patrick Alphonso, Yu Fang, Kelly Lyons, Elena Frank, Srijan Sen

Background: Racial-ethnic minority groups face a disproportionate burden of mental health challenges in the U.S., despite comprising a small proportion of the population. These disparities stem from a lack of community-based interventions, unequal access to evidence-based practices, and insufficient resources to fund health services. One study found that racial minorities are 20%–50% less likely to initiate mental health services and 40%–80% more likely to drop out of treatment prematurely. Given that training physicians experience higher rates of depression, burnout, and suicidal ideation, it is imperative to assess how minority residents engage with mental health 

Authors: Sarah A. Davis, Rohan Sachdeva, Rachel Weingarden, Hannah Swirple, Adam S. Lepley, Zhenke Wu, Taraz Lee, Christopher A. Brooks, Margit Burmeister, Peter F. Bodary, Kenneth M. Kozloff

Background: Mental health is influenced by many factors, including physical activity and sleep. The relationship between these factors in first-year college students is an ongoing research priority. The use of wearable technology and ecological momentary assessments (EMA) through digital platforms offers a means to objectively capture and assess these relationships with high temporal fidelity in real time.
 

Authors: Sarah A. Davis, Mark Wu, Rohan Sachdeva, Rachel Weingarden, Zoe Norman, Hannah Swirple, Adam S. Lepley, Zhenke Wu, Taraz Lee, Christopher A. Brooks, Margit Burmeister, Peter F. Bodary, Kenneth M. Kozloff

Background: The use of wearable technology is rapidly increasing in popularity, specifically among college students. Wearable devices allow students to track physical activity and sleep, offering students a means to self-monitor and reflect upon personal daily habits that may impact their overall performance in the classroom. The YouM study is designed to give students individualized real-time feedback on their habits, including daily physical activity, nightly sleep, and daily Ecological Momentary Assessment (EMA) survey scores through a study-specific dashboard. Accurate use of data from this study requires a better understanding of device use and compliance across the different study variables – physical activity, sleep, and EMA scores.
 

Authors: Authors: Aditya Jalin, Rajnish Kumar, Bengie L. Ortiz, Xiao Cao, Michelle Rozwadowski, Muneesh Tewari, Sung Won Choi

Background: Mobile health technologies offer unprecedented opportunities for longitudinal data collection in clinical research. However, the vast amount and complexity of raw data generated pose significant challenges in data preprocessing, analysis, and reproducibility. Roadmap 2.0, a mobile app for positive wellness-based psychology interventions, uses wearable sensors to collect physiological and psychological data from bone marrow transplant patient-caregiver dyads for up to 120 days post-transplant.

Authors: Shiyu Li, David Conroy

Background: Habit formation is widely viewed as a key to sustained behavior change, yet there is a limited understanding of the pattern(s) or timing of habit formation over time. Characterizing these intraindividual dynamics and individual differences therein, requires intensive longitudinal data. This study aimed to determine whether a uniform habit formation trajectory exists or if different functional forms are needed to characterize changes in habit strength among patients with a history of kidney stones engaged with sipIT, a just-in-time adaptive intervention to promote fluid intake.
 

Authors: Lars G. Fritsche, Elena Frank, Laura Thomas, Amy S. B. Bohnert, Srijan Sen

Background: With the increasing use of mobile technologies in health research, personalized interventions have proven essential for improving patient engagement. In the COMPASS study (Comprehensive Mobile Precision Approach for Scalable Solutions in Mental Health Treatment), we aim to send participants tailored daily messages informed by metrics like sleep quality, mood, and activity levels. To enhance message personalization, we have developed an automated pipeline utilizing large language models (LLMs), particularly GPT-based systems, to generate diverse, contextually appropriate messages.
 

Authors: Margaret Banker, Leyao Zhang, Peter X.K. Song

Background: Biological aging research aims to understand variations in how individuals age biologically and its connection to age-related diseases, with epigenetic age serving as a key biomarker. Wearable devices provide continuous data like accelerometers on physical activity, which may help predict long-term health outcomes by analyzing its relationship with epigenetic age.
 

Authors: Dr. Christopher Brooks, Maikl Awad

Background While performance-tracking technology has advanced significantly in running and cycling, it remains underdeveloped for strength training. Conventional methods for monitoring strength training exercises, such as bench press, have not kept pace with these advancements, leaving a notable gap in effective performance tracking.
 

Authors: Rohan Nanwani, B.S.; Elizabeth Mills, Ph.D.; Amy Bohnert, Ph.D., M.H.S; Srijan Sen, M.D., Ph.D.; Adam Horwitz, Ph.D.

Background: With the proliferation of digital interventions and robust evidence for their effectiveness in reducing mental health symptoms, rigorous evaluation of patient engagement with these digital tools is critical.
 

Authors: Joseph Alphonso, Yu Fang, Kelly Lyons, Elena Frank, Margit Burmeister, Srijan Sen

Background: This research was conducted as part of the Intern Health Study, which uses mobile health data, genetic data and survey data to analyze mental health outcomes in physicians undergoing training. This research seeks to determine whether there is a correlation between genetic chronotype and depression. Generally, the hypothesis is that those genetically predisposed to be “night owls” - wake up later and stay up later - will have higher PHQ-9 (Patient Health Questionnaire - 9) scores during the intern year, than those who are “early birds” - wake up earlier and go to sleep earlier.
 

Authors: Megan Chen, Shriya Anand, Adam Horwitz, Ewa Czyz

Background: Suicide is the 10th leading cause of death in the U.S., representing a significant public health burden. Emergency Departments (EDs) often serve as the first and only clinical contact for individuals at risk for suicide, offering opportunities for targeted continuity of care interventions during the high-risk post-discharge period. Best-practice ED guidelines recommend providing high-risk individuals with brief interventions that include safety planning—identifying coping strategies to mitigate suicidal crises—and follow-up contacts. However, busy EDs often lack the resources to offer these interventions consistently or with fidelity. Innovative approaches that deliver accessible, personalized, and resource-efficient ED-initiated interventions are needed to prevent suicidal behaviors post-discharge.
 

Authors: Allie J. Tracey, Abigail C. Bretzin, Ashley Rettmann, Douglas J. Wiebe, Tracey Covassin

Background: The psychometric properties and usability of concussion symptom checklists and psychological health-related quality of life (PHRQoL) inventories as measured by a novel, mobile ecological momentary assessment (EMA) platform, Recovering Concussion Update on Progression of Symptoms (ReCoUPS) remain unknown.
 

Authors: Shriya Anand, M.A., Megan Chen, B.A., Adam Horwitz, Ph.D.

Background: Depression is a leading cause of disability, with rising rates in the U.S. over the past 25 years (WHO, 2017). College students face high depression rates but often do not seek treatment due to barriers such as time, perceived need, and access (Czyz et al., 2013; Horwitz et al., 2020). Digital mobile health interventions offer new opportunities to overcome these barriers, however the high dropout rates and low utilization suggest the need for a modified, low-burden approach to improve participation.
 

Authors: Victoria Murphy, Sarah Sperry

Background: Sleep disruptions are hypothesized to be an etiological factor in bipolar disorder (BD), and may exacerbate affective, cognitive, and behavioral dysregulation in BD. However, few studies to our knowledge have examined the discrepancy between subjective and objective total sleep time (sleep misperception) in BD and its clinical implications.
 

Authors: Theodor Meingast, Amanda C. Melvin, Sarah Davis, Kenneth M. Kozloff, Alexandra F. DeJong Lempke, Adam S. Lepley

Background: Wearable sensors, such as smartwatches and shoe-mounted inertial measurement units (IMU), have become increasingly popular for monitoring physical activity. These sensors provide valuable spatiotemporal data for runners (e.g. stride length, cadence, ground contact time). Understanding the level of agreement between devices during running exercises at different distances and environments can help guide individuals when selecting the most suitable wearable device.
 

Authors: Megan Johanness, Amanda C. Melvin, Adam Audet, Sairub Naaz, Theodor Meingast, Sneh Shah, Sarah Davis, Kenneth M. Kozloff, Adam S. Lepley

Background: Smartwatches have gained popularity for monitoring heart rate during daily activities and exercise. The rapid advancements in these technologies, both sensor/hardware and algorithm/software, also mean that updated accuracy data is needed to keep pace with the ever-evolving technology and derived metrics provided by these devices. Due to the indirect measurement of heart rate, the accuracy of smartwatch-derived metrics may be affected by exercise type and intensity.
 

Authors: Zubin J Modi, Ashley E. Rahimi, Becky Scherr, Mike Arbit, Hailey Desmond, Abigail Smith, & Eloise Salmon

Background: Mobile health (mHealth) can offer investigators novel approaches to research and data collection while reducing traditional barriers to study participation. Potential benefits to research use of mHealth include representative sample acquisition of rare disease participants, access to external electronic health record (EHR) data, and real-time disease metric tracking. Unfortunately, mHealth initiatives in kidney disease research are limited.
 

Authors: Poonam Purohit, Tristin Smith, Vivian Kurtz, Catherine Klida, Anne Arewasikporn, David A. Williams, Kevin F. Boehnke, Amy S. B. Bohnert, Rachel S. Bergmans

Background: Research studies are increasingly relying on technologies that allow for the collection of large amounts of real-time, complex data across multiple data capture platforms. While this approach allows us to obtain rich data, it raises questions about how to reliably make sense of data that vary in structure and format.
 

Authors: Walter Williamson, Joyce Lee, Irina Gaynanova

Background: The widespread adoption of continuous glucose monitor (CGM) technology has led to a proliferation of CGM data and the adoption of CGM metrics in both clinical and research settings. Unlike curated CGM data from clinical studies, CGM data stored in research data warehouses may have periods of duplicated data uploads from the same patient, often accompanied by time shifts, compromising data quality. These have received limited attention in the literature, leading to the computation of CGM metrics on data "as is" or ad-hoc data processing approaches such as averaging measurements from the same time stamp. Proper identification and correction of data quality issues due to duplication and time shifts is required to ensure the accuracy of resulting CGM metrics. 
 

Authors: Grace Jung, BA; Alexandria R. Schaller, BA; Jenny S. Radesky, M.D.

Background: Smartphone and tablet use has rapidly increased across all life stages, from young children to parents. In response, researchers have investigated the influence of device use on daily life. However, prior literature has heavily relied on self-reports of device usage, which have proven to have low accuracy (Radesky et al., 2020; Yuan et al., 2019).
 

Authors: Sanjana Kandiraju, Katherine Ross, Kelly Lyons, Elena Frank, Srijan Sen

Background: Previous research demonstrates persistent drug abuse in physicians due to occupational stress, particularly in surgeons, anesthesiologists and general practitioners (1). Prior studies suggest that certain demographics predict higher occurrences of substance use disorder. Namely, factors such as age, gender and stress can affect the risk individuals have for developing substance use disorder (2). Research on the prevalence of substance abuse in intern physicians is important because it can help inform research on targeted interventions for at-risk groups and the appropriate timing in which these interventions are placed.
 

Authors: Amani Alkayyali, Max Shtein, David Lipps

Background: The shoulder’s complexity, with six degrees of freedom, makes measuring glenohumeral joint movement challenging. Monitoring shoulder kinematics is crucial for injury prevention, rehabilitation, and optimizing performance, especially restoring internal and external rotation for athletes and post-surgery patients. However, quantifying shoulder rotation in clinical settings is difficult due to the large, expensive setups required for motion capture and the precise calibration needed for inertial measurement units.
 

Authors: Rui Nie, Zheshi Zheng, Peter Song

Background: Type II diabetes (T2DM) is a chronic condition that affects how the body regulates blood sugar (glucose). As the prevalence of this disease continues to rise, understanding its underlying mechanisms, improving treatments, and preventing its onset has become increasingly important. T2DM is often linked to lifestyle factors, which motivates us to analyze data from the All-of-Us program. This program is launched by the NIH, and integrates clinical, environmental, and lifestyle information from a diverse population. Specifically, we focus on Fitbit data from more than 15,620 participants, which captures daily active zone minutes, through wearable devices.
 

Authors: Abigail Bretzin, Jason Goldstick, Allie Tracey, Douglas Wiebe

Background: Concussion is a common injury among pediatric patients, and concussion patients present to their healthcare provider with a vast amount and diversity of symptoms (i.e., somatic, vestibular-ocular, cognitive, sleep, emotional) on their first visit. Research to date proposes an important prognostic indicator includes early symptom presentation, however healthcare providers often repeat symptom assessments at each visit to intervene with treatment and monitor recovery. Remote daily symptom monitoring via SMS texting and paired dashboard may help healthcare providers monitor recovery between visits.
 

Authors: Swaraj Bose, Walter Dempsey

Background: Accurately estimating individual treatment effects (ITEs) across multiple decision points is crucial for personalized decision-making in dynamic environments such as healthcare, adaptive interventions, and financial forecasting. Previous work has primarily focused on constructing prediction bands for ITEs using cross-sectional data or predicting non-causal longitudinal estimands, both relying on assumptions of exchangeability.
 

Authors: John Sohn, Hoonyong Lee, SangHyun Lee

Background: Identifying and eliminating slip, trip, and fall (STF) hazards is a fundamental strategy for preventing fall accidents. To detect STF hazards, current practices rely on manual inspection, which requires substantial time and resources to find and fix the hazards. To address this limitation, the authors propose wearable sensors and a collective sensing-based approach to detect STF hazards. Specifically, this approach utilizes the potential of wearable motion sensors and biosensors to continuously monitor an individual's physical and physiological responses that might abnormally change due to STF hazard exposure.
 

Authors: Saagnik Sen Dasgupta, Abigail C. Bretzin, Matt Lorenz, James T. Eckner

Background: Concussion is a common injury among pediatric patients, and while understanding recovery outcomes and determining interventions is often limited to subjective symptom reporting. Whereas, identifying opportunities to monitor patients between visits may yield vital information that can inform clinical decisions and speed recovery. In addition, recovery from injury during sleep periods is likely important, but widely understudied.
 

Authors: Raphael Jeong Hin Chin, Rahul Ladhani

Background: Mobile app-based games have emerged as promising tools for engaging children in health education, particularly in resource-constrained settings. Fooya! is a clinically proven mobile application designed to promote healthy habits among children through engaging gameplay. While previous studies have demonstrated its effectiveness in improving nutritional choices, the specific gameplay patterns that most effectively enhance knowledge of infectious disease prevention remain unexplored. 
 

Authors: Henrike Florusbosch, Ph.D., e-HAIL program; Rada Mihalcea, Ph.D., Electrical Engineering and Computer Science, College of Engineering; Akbar Waljee, M.D., Learning Health Sciences & Internal Medicine, U-M Medical School
 

Abstract: This poster will present an overview of the e-Health and Artificial Intelligence (e-HAIL) program at U-M. Established in May 2022 as a joint initiative between the Office of Research in the Medical School and the Associate Dean for Research in the College of Engineering, e-HAIL’s mission is to make the University of Michigan a premier hub for e-health & AI research to improve health using technology. The program focuses on fostering new multidisciplinary collaborations between AI and Health experts, supporting research/grant development to secure extramural funding, and providing infrastructure to enable innovations in healthcare and AI/ML methodologies. The goal is to develop and deploy AI-based tools for improved health outcomes. If you're a faculty member (interested in) working at the intersection of AI and Health, our program manager is always happy to learn more about your research and how e-HAIL might be able to support it.