U-M's Progress and Resources in Utilizing Mobile Technologies in Health Research | DIGIT-MI
I am Kathy Goldstein. I'm a neurologist in the Sleep Disorder Center here at Michigan Medicine. And my scholarly work had landed in the wearable tracking of sleep and modeling wearable signal to predict sleep in circadian rhythms.
And so I got hooked in with doctor Sen and the Eisenberg Family Depression Center. And I'm now that faculty leader at the Mobile Technologies Cre. And this session, especially if you're new to digit MI, this is kind of a typical for most of the sessions. But what we're going to talk about today is how the mobile technology core has progressed and some of the resources we've developed ourselves, but also identified throughout University of Michigan in helping research teams utilize mobile technologies in their health research. And our two main goals, keep it simple here, so we wanted to reduce friction for investigators that wanted to use mobile tech in their research. And most importantly, we didn't want to duplicate anything that already existed on campus.
So most of you know this, there's a lot of benefits to incorporating mobile technologies into your research. So the way the measurement works, it allows for simultaneous assessment of different objective parameters. So for example, heart rate variability, along with sleep, along with activity, and also because these devices are associated very easily with mobile apps. Subjects can provide self report measures as well.
The data collection occurs in a free living ambulatory condition, and it's over long term. It's passive, so reduces burden on study subjects, and we can be provided pretty easily with contexts around the situation the data was collected. The characteristics of using mobile tech in research allows you to scale your work and also reach people that might not typically be involved in research. And kind of more things on the horizon is that we can now digitally phenotype people with this data.
We might be able to come up with new hypotheses about things that we couldn't usually observe in the ambulatory state, and we have new ways to design interventions that are actually going to take place in free living conditions. But what about challenges? So we noticed when we started a core, we were really unclear what resources University of Michigan had even available. We needed a more coordinated effort, coordinated institutional support. And then there was a lot of heterogeneity. So we needed standardization and also some resources devoted to infrastructure to really make this type of research effective.
So the mobile Technologies Corps started developing resources to address some of these challenges. And our resources, we can kind of break these down into our resource library, as well as direct research services. So starting with the resource library, we have our knowledge base, and the knowledge base is fulled with full of articles that are going to help people conduct their research with wearables and other mobile technologies.
And they can cover a variety of things. So some of them are just about the resources we have at UOM for this kind of work, how to get started in incorporating mobile technologies into your research, how to use secondary datasets that were projects that used mobile technologies to acquire their data and best practices in this kind of work. And we're hopeful that all of you who have important expertise in these areas will collaborate with us so we can develop new articles as these have been very helpful and actually accessed by people across the world. So here's just an example of one.
This is the essential guide to sharing code and biomedical research because this has now become very, very important. And so this gives a lot of the whys and the how tos in this article. This is also where we provide documentation for our code, and I'll talk about our code repository in a moment. So we do have a GitHub code repository, and you can see some of our project listing there.
This contains code that has been developed through the mobile technologies core, as well as through the UM M mobile tech community at large. So people have been great at sharing this code with us, as well as some external repositories, we provide links to those because there are some available that have code for the processing of mobile acquired data. Again, we don't want to duplicate anything that already exists. So This allows you to share your code if you don't already have a code repository associated with your lab, we're happy to host your code for you. There's a DOI number that comes with it, so you can see who's using your code when they go to publish.
We have a read me file, and this meets NIH requirements for data management and reproducibility. And it's maintained by our mobile data experts network Affinity Group, and it's been really really useful, so I hope everybody can check that out. So here is just one example. This is actually the article about the code. So this actually lives in the knowledge base.
But this is code that we developed for Helen Burgess's lab. She was collecting sleep data with the Fitbit. But in sleep, we get very concerned about how we incorporate the objective sleep data from Fitbit and similar devices with the intent to sleep. So where is the period that the patient is actually trying to sleep as opposed to what is the period of time that they have active or excuse me, resting wakefulness, where they're sedentary scrolling through their phone on Netflix, whatever. So what this code does is it automatically harmonizes the patient's self report of when they are actually about to start attempting sleep, and when they are done attempting sleep, and this harmonizes that data with the objective sleep data from Fitbit via Fitta base, and then calculates aggregate sleep parameters that are what we use as covariates and outcomes in our research. And this allows for more accurate measurement of the wearable sleep parameters like sleep onset, how long it takes you to fall asleep once you try, sleep offset, total sleep time, sleep deficiency, et cetera.
But this code allows this data cleaning to get those accurate measures to occur in a more scalable way because it's now automated. So that was this was an exciting win for us. Also included in our resource library, our whole collection of videos from different lectures, including from this group, Digit MI.
We have 26 videos now, over 2000 views. They've been on a variety of different things, methods, how tools, tools and resources that are available throughout different departments at University of Michigan. So and these are through YouTube.
We have a questions and discussions area of our resource library. This allows you to ask other people throughout the University of Michigan Mobile Tech Community. And the great thing about this, it's searchable, and likely if you're having a problem, somebody else has either had it or has had a similar issue, and you can figure out what people have done to address this.
Okay, so moving from our resource library, which is educational manual, et cetera, to the direct kind of live research services that we have. So we do, through our mobile technologies core website, have a place that you can request mobile study research consults, and we really kind of see the entire mobile technologies research life cycle within these consults. So everything from selecting your wearable device or your type of technology, what kind of mobile data domains do you need to get to your actual research question, what methodology should be used. We've helped people at data pipeline development and data processing flow.
We have the ability to provide sample protocols, consent, and other study documents, the automations, which was stem that I just showed you, that stemmed from a research, consult, other guidance and assistance in data cleaning. Assisting with regulatory and compliance issues that are directly related to research with mobile technologies. And also, a lot of times, you know, we find that one of our greatest benefits is connecting people with others that could provide the service or provide resources. So we're oftentimes sending other people to other U of M units if there are things that are beyond our scope. We do have also have a database of University of Michigan researchers and the areas and the products that they have expertise in. We can connect people for mentorship questions and collaboration.
And we have something new coming in the pipeline. We're going to have Hub Med keyword searches. So that's in partnership with the Top Men Science Health library so that we have different relevant mobile technology research keyword searches kind of already canned and available for people to take a look at. Now, what about coordinated institutional support? How do we make University of Michigan the best place to get research done with mobile technologies? So there's a few different ways that we've done this, a large part of this, as you can imagine with finding all these individuals and building community, our metric symposium, working on a project for clinical trial management, and also working on a project for texting and research, as well as some infrastructure development down the road? So our UM M mobile technologies communities started, gosh, it's almost three years ago now when we formed the mobile technologies Cre, and we began soon after that, offering services. We started our mobile data experts network. We digit MI, which has been around a long time.
We subsumed that under the MTC since that was in the same content area in January of 2023, we formed the mobile coordinators network for our study coordinators with expertise in mobile technology And then this past summer. So a little over a year ago, we formed the mobile technologies research and innovation collaborative of Metric because individuals who are interested in using mobile technologies research, we found go far beyond the Depression center and far beyond just the medical campus, so we wanted to make sure everyone was involved, and we formed metric to use as kind of an umbrella for that. In November of 2023, we have our first metric symposium, and we have our second annual coming up shortly. So this is kind of how we imagine the way the organization of these groups work.
So with metrics, the units that the different members come, they can either be engaged in research, or they support research, or they develop research methods that utilize mobile technology, and it's comprised of both a faculty leaders and administrators. So it's a very broad group. And then within metric, we've housed our three affinity groups, so Digit MI, We have speakers who lead mobile study research and educate and inspire the global community, that's this group, which is often except for today, going to have presentations about specific research projects and the lessons learned and challenges posed by those. We have faculty staff and learners who attend this. MDN, again, our mobile data experts network.
These are data analysts who are sharing methods to make mobile data easier to process, access, and the data collection cleaning and analysis reproducible. So this is data analysts. And then our mobile coordinators network, who study study share study protocols, best practices, consent methods. And so through all these different areas, these form really nice groups across that mobile technologies research life cycle.
And the thing about metric, it's kind of confusing, you know, as many of these hierarchies are at UM, but it's not we don't have organizational standing or resources. This is just a group that brings everybody together, and then the units have their own resources. And our metric symposium will be on Friday, November 1. You can see the links to register and the agenda, and if you want to submit a poster there. We have some excellent keynotes, as well as panels that were really exciting and lively last year regarding kind of resources and conduct of mobile technologies research.
And 100% of our attendees last year when they were surveyed, they would recommend this to a colleague. So I hope to see you guys all there. We were also lucky enough to have all of these different groups sponsor metrics symposium. So for those of you within these groups who are here right now, we're greatly appreciative. And what's really beautiful about this is this allows us to conduct this conference without registration, which is great for students and our butting mobile technology researchers of the future.
Okay, our clinic trial management system project. So the goals of that project. So when people do their mobile technologies research, as most of you know here, you need a software platform to ingest that data, manage the data and the study activities, and then to act as a conduit to get that data out. And people were using different vendors. Some had more robust and collaborative relationships with the University of Michigan. So we really wanted to form a group of these vendors that could be people's go to when they were looking for these software platforms for their mobile technology research.
So we wanted to select these. They also had to meet the needs of UFM researchers. So we created a feature and technical requirement kind of wish list. And this also, because of the procurement process, by having this group of vendors to streamline the selection and the approval processes. So We're hopeful that with the selected vendors, we're going to meet the needs of different groups of researchers, whether you're junior doing a small study, or whether this is a large scale study for more seasoned investigators, and these will be, you know, cost appropriate, and they can be generalizable to do other things. And we're hopeful that this is going to solve that challenge.
So we've already evaluated at least 20 study platforms, and we're going to select the ones so that we can expedite Information assurance approval, so that's already done for our researchers. And then we will have documentation about all the features of these different vendor platforms, and we'll share those. They actually, the manufacturers came to demo, and we have videos of them demoing their products. So again, you can make a more guided decision on what software to buy. We're using this to facilitate a master service agreement, and what this also allows us to do, is when we develop data pipelines, we are developing data pipelines that work with the selected platforms. And when we're developing code for data processing and cleaning, again, they work with how the data is coming from these platforms.
Also, working on dashboards with these because this has been one of the more difficult things in research is having a dashboard that study team facing that is going to meet the teams need. You can see the list of vendors under consideration. And the mobile coordinator network, again, this is Aimed Ater Study coordinators, is going to have a meeting about this at their September 25 session. Another big problem we identified among our researchers using mobile technology is that the options for texting and research were not working well. So we submitted a request for an institutionally supported SMS service through the HITS Das portal, which I actually didn't know existed up to then.
And we got a huge response. Within days, the idea received all these up votes. So now HITS has considered this a priority. They're investigating options.
They're surveying people who were involved on the portal and meeting with stakeholders. And there's a link there where you can add your either issues with texting and research here or what you think is needed. And Victoria, she always says this to me in our meetings. The answer is no until you ask the question. So we're hopeful that we move the needle on this.
With a texting vendor, obviously, again, we have the information assurance issue. We want to pick somebody that will meet those requirements. This should be something that's, you know, almost turnkey. We don't want study teams to have to hire more specific IT staff to use a texting feature, automation of text replies, so those don't have to be done manually. We want access to all of the SMS data. Through an API, ability to send bulk text reminders, ability to use only one phone number for the whole team, so there can be two way texting as well.
And then ideally we would have integration with some of the platforms we use. So Redcap, my data helps Sita base and PAL, so that this is a seamless integration with the texting and the wearable vendor platform that the vendor provides for the wearable data management. Another thing that we found is a gap in the support for mobile technologies research here is there are steps in the flow of data that become very difficult for our investigators and they'll do a study. They know the device they want to use, they know the measures they want to collect, they even get this that vendor third party platform to collect the data. But then things come a little bit more difficult in moving the data into the U ofM environment.
Cleaning that data and having appropriate tools that are specific for mobile technology data for analysis and visualization, and then data storage. So to address that, we're developing a common data model for mobile data, as well as a data pipeline. Those are two projects that are just at the beginning. So our common data model project, we want to develop a standardized data model for mobile data and all of the different content areas and all of the different physiological parameters, and then develop code and associated tools to support those.
And we think that this will increase efficiency number one, but also rigor and reproducibility far beyond University of Michigan because these parameters are being used in different ways across research. And these are the devices that we have obtained to pull the data from. And we are acquiring experts in different content areas. So with the cardiac data, the activity data, the glucose data, the sleep data to help us come up with this common data model And we will using those devices, identify measures that are clinically relevant. So we want to use things that are going to help translate from this mobile technology research to important clinical care.
We'll have data dictionaries for the standardized data model. We'll have a data template. There'll be open source algorithms to use, and we're also going to develop data transformation code for after the collection, share this code through open source licensing, and we want to publish our results, so other people can use this common data model for themselves, and we hope it becomes a research standard. The data pipeline project, the reason we initiated this. So currently, again, after that data ingestion by that third party software platform, a lot of times that data was then being manually downloaded and transformed for analysis and with tools that are not necessarily equipped for mobile data, this was inefficient, this was costly. So our goal is that we need an enterprise grade data pipeline tool, as well as a tool that can transform raw data into structured formats in a modular and maintainable manner.
We're also going to need dedicated servers for this and a research integration analyst, but we would like to make a data pipeline that is generalizable to different projects across the university. So we've had a lot of people say, Mobile Technologies Core, you're in the Depression center. I'm not a mental health researcher. Do I really need your help? You know, we would say that use of Mbal technologies research really really does facilitate the integration of mental health components into research that might not have typically included them.
And we've had consults from everywhere and from orthopedics and a chronology, GI, OB. So we really are finding that these are getting incorporated across different domains. And here's just our aim, but I think I got at that with everything I've discussed and our team that is absolutely amazing,
2024-09-17 23:53
