Jordan A. Carlson, PhD, MA
Director, Research Center for Children's Healthy Lifestyles and Nutrition; Professor of Pediatrics, University of Missouri-Kansas City School of Medicine; Research Assistant Professor of Pediatrics, University of Kansas School of Medicine
Full BiographyMark Hoffman, PhD
Chief Research Information Officer; Professor of Pediatrics, University of Missouri-Kansas City School of Medicine; Research Professor of Pediatrics, University of Kansas School of Medicine
Full BiographyJ. Steven Leeder, PharmD, PhD
Marion Merrell Dow Endowed Chair in Pediatric Precision Therapeutics; Deputy Director, Children's Mercy Research Institute; Associate Chair for Research, Department of Pediatrics; Professor of Pediatrics, University of Missouri-Kansas City School of Medicine; Research Professor of Pediatrics, University of Kansas School of Medicine
Full BiographyMark A. Clements, MD, PhD
Rick & Cathy Baier Family Endowed Chair in Endocrinology; Director of Clinical Research, Endocrinology; Medical Director, Pediatric Clinical Research Unit; Professor of Pediatrics, University of Missouri-Kansas City School of Medicine; Clinical Assistant Professor of Pediatrics, University of Kansas School of Medicine
Full Biography
One of the strengths of the Children’s Mercy Research Institute (CMRI) lies in its integrated approach to research and support for multidisciplinary labs and research groups. The Research Informatics and Data Science team led by Mark Hoffman, PhD, Chief Research Information Officer, is an example of a group that collaborates with researchers throughout the CMRI. The team provides a variety of services, including software design, technology review, data warehousing, high-performance computing services, and project consultation.
An inhouse informatics team that can develop and deploy software solutions across research groups provides an economy of scale. “We listen for common requirements across multiple projects,” said Dr. Hoffman. “We can do a foundational development, and with some tweaks, deploy again. Taking that platform approach has enabled us to onboard new projects much faster at a lower cost.”
The Research Software Engineering team, part of Research Informatics and Data Science, worked with pioneering principal investigators (PIs), Jordan Carlson, PhD, MA, Director, Community Engaged Research, Center for Children’s Healthy Lifestyles & Nutrition, and Mark Clements, MD, PhD, Director, Rising T1DE Alliance Initiative and clinical research for Endocrinology, to incorporate wearable health and fitness devices into their studies. Dr. Carlson and Dr. Clements had unique specifications and different uses for the wearable devices, but the projects also had common requirements.
By identifying the foundational requirements for both projects, the software engineering team developed a scalable platform for integrating wearable devices into projects moving forward, including a recent project led by J. Steven Leeder, PharmD, PhD, CMRI Interim Executive Director, and Catherine Koertje, MD, Project Manager, both from Clinical Pharmacology, Toxicology and Therapeutic Innovation.
In an early initiative to integrate wearable devices into CMRI research, Dr. Carlson wanted to not only use wearable fitness devices in his research interventions but also build a system to communicate back and forth with the research participants to encourage physical activity. He wanted to base individual communications on the data collected from the fitness devices. “At that point, we only had one-way communications, like informing participants about upcoming appointments. Dr. Carlson wanted to take it a step further to communicate in ‘real time’ with study participants,” said Harpreet Singh Gill, Manager, Research Software Engineering. “It was a complicated project because it took custom decision logic. The decision trees about how each decision is made were six to seven layers deep.”
The system that the informatics team developed has the capability to immediately process incoming device data and then automatically assess where each participant stands in a study. “An example goal for a participant was 5,000 steps every day,” said Gill. “If we detected that a participant needed additional support to meet this goal, a notification based on the decision tree would go out to the participant.” Notifications are sent via text to a participant’s personal mobile phone, using theory-based behavior-change content, such as helping identify and problem solve barriers to physical activity. Participants can respond to the text, and those responses would trigger additional messages from the system and/or notifications to the study team. This is an entirely automated process.
The Research Software Engineering team has programmed and built the software around all our decision trees. Over time, the team has made the system more sophisticated to accommodate preferences we hear from participants. The team even built a web-based portal where we can access the data. The system keeps getting better and better and the work has expanded.
“The Research Software Engineering team has programmed and built the software around all our decision trees,” said Dr. Carlson. “Over time, the team has made the system more sophisticated to accommodate preferences we hear from participants. The team even built a web-based portal where we can access the data. The system keeps getting better and better and the work has expanded.”
The Research Informatics and Data Science team launched the web-based Health Device Portal (HDP) in 2023. Through HDP, researchers can manage data collection from health devices such as fitness trackers, scales, and glucometers. HDP provides preconfigured reports and a self-service dashboard to access study data. CMRI research teams can grant portal access to their external research partners. Currently, the portal supports Garmin® and Fitbit®. “Before HDP, our team was involved heavily in setting up a study and managing access to these technical resources,” said Gill. “The portal was a step forward to make this a self-service program with ease of access.”
Dr. Carlson and his team have now integrated wearables into several projects to support physical activity in community members from lower-income, predominantly Black communities who often experience health disparities. In the Active KC study, the goal was to encourage increases in physical activity. His team provided participants with wearable devices from Garmin®. The intervention included participant-tailored text messages (e.g., goal setting and monitoring) that were automated by the Research Software Engineering team and used real-time data from HDP to provide customized content to each participant.
Dr. Clements is also pioneering the use of wearable devices in his Type 1 diabetes (T1D) research. He sees value in wearable device data from both the research and clinical perspective. “We continue to be interested in wearable data because we think they can inform population health insights that can drive innovations and improvements in care,” said Dr. Clements. "Being able to work with the Research Informatics team enables this kind of work because we wouldn't be able to do it if we couldn't design new software in the way we need it to be implemented. When we work with existing third-party software solutions, we're often trying to fit a square peg into a round hole.”
Dr. Clements and his research team are using the new HDP for the Garmin® PACT (Physical Activity Tracker) study, which is part of the larger project, Rising T1DE Alliance, led by Dr. Clements. Project collaborators include Ryan McDonough, DO, Endocrinology and Health Informatics and Technology; Susana Patton, PhD, Nemours Children’s Health, and Sanjeev Mehta, MD, MPH, Joslin Diabetes Center. The goal of the PACT study is to see if wearing the Garmin® physical activity tracker increases activity and control over T1D. A secondary goal is to see if wearing the Garmin® tracker changes how patients feel about managing their diabetes. “Physical activity is one of the pillars of diabetes self-management,” said Dr. Clement. “Remarkably, it gets the least attention in diabetes care. We have thought about how we can develop interventions that leverage Garmin® data so that we could try to nudge participants toward more physical activity.”
Dr. Clements began integrating Garmin® data into software systems that support research in 2017. In the project, “Type 1 Diabetes and Exercise Initiative (T1DEXI)”, Dr. Clements collaborated with researchers across the U.S. to study the effect of exercise in T1D. The team’s findings appeared in a 2023 Diabetes Care article. “Our integration of the Garmin® data in the DEXI study was our training ground, mostly to understand how to wrangle the data,” he said. “The informatics team used this early work to inform their current approach. We worked with them, and they learned what did and did not work.”
Being able to work with the Research Informatics team enables this kind of work because we wouldn't be able to do it if we couldn't design new software in the way we need it to be implemented.
The Research Informatics team also integrates other types of health devices. “When we talk about wearable devices, it’s across the board — not just fitness devices but also devices like glucometers,” said Gill. “We are working with Dr. Clements’ team, developing data collection for a project with a scale multiple times more than what we get from Garmin® and Fitbit®. Dr. Clements noted that the project will impact all 2,700 patients in the diabetes center and support more than 200 devices.
Dr. Koertje and Dr. Leeder lead the GOLDILOKs® PRISM study and have integrated wearable fitness devices into the study. The team is collaborating with researchers at the Mayo Clinic who developed a machine learning model to predict whether a teen with depression will respond to antidepressants within a few weeks of treatment. The goal of the GOLDILOKs® PRISM study is to match patients with an effective treatment as soon as possible. A secondary goal is to refine the Mayo Clinic model by identifying individual factors (physiological, genomic, metabolomic) that may affect whether the antidepressant fluoxetine (Prozac®) will improve depression symptoms. To capture physiological data, study participants wear a Garmin® device.
“Physiological data collected includes heart rate, sleep, and physical activity,” said Dr. Koertje. “We are interested in whether variations in those measures are associated with depression. We are looking for meaningful patterns in how physiological factors change early during treatment with an antidepressant like fluoxetine (Prozac®) and how the factors respond with or relate to treatment outcome. The wearable devices provide noninvasive, objective data that is easy for participants to do.”
We listen for common requirements across multiple projects. We can do a foundational development, and with some tweaks, deploy again. Taking that platform approach has enabled us to onboard new projects much faster at a lower cost.
The GOLDILOKs® PRISM team uses the new HDP, but they also requested a customized report to see how participants used their watches in real-time. This allows Dr. Koertje to immediately troubleshoot if someone isn’t wearing the fitness device and thus not gathering data. A bonus from creating the customized report is that the Research Software Engineering team can easily adapt it for other research projects.
“Harpreet’s team was able to deliver exactly what we were looking for, including with the customized report,” said Dr. Koertje. “We look forward to seeing what the data shows and if we can find some meaningful patterns behind this. I think the potential is there.”
Dr. Leeder added, “I don't understand all the processes involved in collecting the data, but fortunately, we don't need to because we have people here who are experts in that area and are committed to making sure research teams have the best tools for their studies. We would not be able to include wearables in our study if it weren't for the informatics team.”
“Collaborating with Harpreet’s team has enabled us to very feasibly use cutting-edge technology in our research and interventions,” said Dr. Carlson. He noted that wearable technologies give participants flexibility over when to engage and are critical for reducing burden on participants. At the same time, he said, “It helps us maintain regular, ongoing support for participating families.”
“Having Harpreet’s team’s capability inhouse adds significant value. His team understand our needs and the needs of our patients and families,” he adds. “They are extremely flexible. They don't just offer us a core set of services. They provide customization and keep building on the original system. This partnership is essential for the feasibility of our work. Grant reviewers can see that we have this proven, well-developed infrastructure and experienced informatics team inhouse, which substantially strengthens our grant applications.”