Innovations in Health

Through Artificial Intelligence

A LETTER FROM OUR CONVENERS

The e-Health and Artificial Intelligence (e-HAIL) program was established at the start of FY22, anticipating the increased importance of Artificial Intelligence (AI) in healthcare in terms of research and the potential for methodological and clinical innovation. Our vision is to enable technological breakthroughs in core AI methodologies that mitigate against health inequities and positively impact the health of individuals and populations, within healthcare systems and beyond, both in the US and globally. Our overarching goal is to develop and deploy AI-based tools for improved health outcomes in line with U-M’s Vision 2034, which emphasizes human health and well-being along with a core commitment to research and AI.

Looking back on e-HAIL’s third year, we are pleased to note that the program has emerged as a critical node in the rapidly evolving health research ecosystem at U-M as it has been transformed by advancements in AI. e-HAIL fills the critical gaps of connecting core AI faculty to clinical experts in areas of critical healthcare needs, identifying promising areas for collaboration in “AI for health” research. Success in this space increasingly hinges on the multidisciplinary strength within research teams, the deployment and initial evaluation of models in clinical contexts, and access to multi-site health data. e-HAIL monitors and responds to trends in funding opportunities and identifies and solves barriers in order to increase faculty success in securing large-scale/MPI grants to carry out transformative AI for health research.

e-HAIL remains steadfast in its commitment to its core mission of research and faculty support through research/grant development, community/collaboration building, and shared resources. Our 2024 Impact Report underscores how e-HAIL's services are responsive to faculty needs and showcases e-HAIL member grants, projects, and other successes in the past year. We hope it motivates you to become part of e-HAIL’s dynamic community of faculty researchers who collaborate to advance AI and health at the University of Michigan and beyond.

Core Support Area Metrics

What sets Hossam apart is his commitment to thoroughly understanding the goals of the grant and going above and beyond to ensure that, when submitted, the proposal is the best it can be. His dedication always translates into concrete and often significant contributions that consistently enhance the overall strength of the grants he works on.

From the 2023 successful nomination of Hossam Abouzahr, e-HAIL Grant Writer, for the Michigan Medicine Office of Research “Rookie Rockstar Award”

I appreciated the opportunity to present at an e-HAIL session because I received insightful suggestions about both experimental design and computational modeling approaches from colleagues. I also connected with two new collaborators who met with me one-on-one after the talk.

Working with Ye Chan Kim, e-HAIL Programmer, we've made significant progress on a decision support tool for developing National Essential Diagnostics Lists, surpassing all previous attempts with other partners. Ye Chan's thoughtfulness, efficiency, and expertise have been instrumental in creating a useful tool that will soon be available, making our extensive diagnostic database accessible and sustainable.

We greatly appreciated the e-HAIL student support funding because we hoped to explore some new ideas and gather preliminary data, but didn't have the bandwidth alone. We believe this will help our future grant proposals immensely.

New External Awards Spotlight

Detecting dynamic fluctuations in emotion, mood, and functioning: A digital phenotypic approach to clinical monitoring in bipolar disorder

NIH funded

e-HAIL members Emily Mower Provost, Sarah Sperry, and key collaborator Melvin McInnis were awarded an R01 grant from the National Institute of Mental Health (NIMH) for a new study that uses Machine Learning to monitor changes in emotions based on speech. Part of the preliminary work for the application was supported through the e-HAIL Student Summer Support Program, which can be used to hire students at any level.

As academic researchers, one of our mandates is to train the next generation of researchers, providing guidance, training, and opportunities. However, due to the complexity of many of our problem domains it can be daunting to figure out how to productively engage with first-time undergraduate researchers. We have been including them on our projects to provide them with the research opportunities that will start their research career.

Leveraging Causal Mechanisms to Build More Robust Machine Learning Models

NSF funded

e-HAIL member Maggie Makar, working with the Director of the Chronic Pain and Fatigue Research Center Daniel Clauw, received support from e-HAIL’s grant writer and was awarded an NSF Career Award to develop more robust machine learning models backed by causal reasoning.

Hossam was incredibly helpful in helping me edit my NSF CAREER grant proposal, making the entire process smooth and successful. He had an exceptional ability to quickly provide feedback on my write-up no matter how late I sent him my draft, which was immensely helpful in the rush to get the proposal submitted to the sponsor.

Increasing Diversity in Biomedical Research

NIH funded

eHAIL member Cornelius James studies barriers to using mHealth tools and the impact of AI/ML on clinical reasoning and medical education curricula. Under the mentorship of eHAIL members Brahmajee Nallamothu and Mike Dorsch and with support from the eHAIL grant writer, he was awarded an NIH Administrative Supplement Award to study patient attitudes toward a just-in-time adaptive intervention (JITAI) mHealth tool for daily sodium intake.

“We don’t know very much about features of this mHealth app that will make clinicians more likely to prescribe it, or the features that will make patients more likely to begin and continue to use it. That’s where my research comes in. It will contribute to our knowledge base about digital health tools, where we currently understand little about how clinicians and patients think of utilizing these tools in their routine care. My goal is to understand and ultimately address barriers to clinicians and patients using novel technologies.”

E-HAIL Projects in Process

Organ Allocation: FROM SPECIFIC AIMS SPRINT TO MULTI-FACETED COLLABORATION FOR DIRECT CLINICAL IMPACT

e-HAIL members Mariel Lavieri, Ji Zhu, and Danielle Haakinson are leading a project to create a Machine Learning model to allocate kidneys for transplants in a manner that improves matching and reduces wait times to save lives and improve quality of life for patients. The multi-team collaboration started with an e-HAIL Specific Aims sprint to brainstorm ideas in response to the real-world challenge of kidney donor allocation.

It has been incredibly rewarding to work with bright investigators and students through the KTAlgorithm project supported by e-HAIL to help us find solutions to expand access and minimize non-utilization of kidney grafts for transplantation. There is significant need for these tools that allow us to better prognosticate which kidneys are suitable to take a risk on as well as which patients would most benefit from these offers.

Stick-figure cameras preserving patient privacy

A new camera designed by CSE researchers, including PhD student Yasha Iravantchi and e-HAIL member Alanson Sample, protects users' privacy in photos and videos taken by smart devices by turning them into stick figures.

Supported in part by the e-HAIL Summer Student Support Program, our team has been working on privacy-preserved sensing (including cameras and microphones). Practical uses so far in the health domain include detecting the frequency and quantity of voiding events in a patient's bathroom after surgery, and we are eager to connect with additional medical partners that could use privacy-preserved sensing to help advance their own research needs.

Data Science in East Africa: AI for health can be particularly useful in global/LMICs

e-HAIL members Geoffrey Siwo, Akbar Waljee, Ulysses Balis, Arvind Rao, and others are working on research with Kenyan partners that uses innovative AI-based technologies to address health challenges and disparities in low- and middle-income countries.

The responsible development and implementation of data-driven solutions is key to addressing Africa's health challenges. Our work in close collaborations with leading researchers across the continent is leveraging rapid advances in biomedical and computational sciences to address health challenges, ranging from drug discovery to population health.

Pandemic Preparedness: The Promise of AI for Health – Pandemic Prevention/Responses

The Computing Community Consortium (CCC), for which e-HAIL convener Rada Mihalcea serves as a Council member, held a visioning workshop on the “Future of Pandemic Response and Prevention” in September 2023 in Ann Arbor, Michigan. Organized by the CCC Council’s Computational Challenges in Healthcare Task Force, the workshop resulted in a report on how computing, and particularly AI, can prevent and mitigate the effects of pandemics.

The COVID-19 pandemic laid bare major gaps in our country’s healthcare system and its technological toolkit. Computer scientists are in a unique position to develop solutions for many of these inefficiencies, but we have a lot of work to do. The CCC workshop held in Ann Arbor, attended by numerous e-HAIL members, was an important first step in that direction.