Despite the high prevalence of diabetes, managing the disease effectively is challenging. Continuous glucose monitoring systems have been used to measure glucose, which is a key indicator in diabetes. Because glucose is affected by the nutrients in the food people eat, it can change quickly and dramatically causing life threatening issues. Therefore, understanding what people with diabetes eat has been a key target of the field, but measuring food intake in the wild has been elusive. This project flips food measuring on its head by using changes in glucose to understand what has been eaten.  

The interdisciplinary team studies food intake to understand individual glucose responses and extrapolates these data to create a model of what people are eating. Doing this work requires advances in Machine Learning (ML) that let the team parse this difficult and ever-changing data. By understanding what people are eating, the biomedical community can create more effective interventions. The project’s early work has shown the feasibility of the approach for understanding carbohydrates, fat and proteins. The research team has already published at multiple conferences and in journals and supported the contributions from multiple graduate students. 

“The project is novel and potentially transformative. It depends on the crucial assumption that the glucose data can be used to accurately predict macronutrient composition, but it has high potential impact. Being a high-risk high-payoff project, it inverses metabolic models to predict macronutrients from the glucose signatures,” said NSF Program Director Wei Ding.  

Researchers at Arizona State University are investigating substances found in sweat and blood, such as protein, which can detect health and disease biomarkers. Researchers will measure changes in these proteins, to develop an inexpensive and disposable patch sensor that can be worn on the skin. Ideally, these sensors, which are now being developed for in-home COVID-19 assessment, can become as widely used as a thermometer found in most home medicine cabinets. 

This project aims to create a miniature electronic sensing platform that can be combined with state-of-the-art biomarker proteomic detection technology to diagnose and monitor multiple diseases with medical laboratory level sensitivity. This proposal presents a new approach combining low-cost commercial display technology (found in your TV, computer monitor, or cell phone) with protein microarray printing technology to fabricate a low-cost, disposable sensor for more effective self-management of patient health care in the home or in other non-clinical settings.  

“This project is a key for individual self-management and since each pixel can look for a different biomarker, each patch can provide feedback for a range of conditions, in many different environments,” noted NSF Program Director Sylvia Spengler.  

Autism Spectrum Disorder is a broad-spectrum developmental disability, with a growing prevalence rate that has seen a four-fold increase over the last 5 years (1 in 44 children, Centers for Disease Control and Prevention 2021). Early intervention involving family and caregivers is very important to help reduce the burden of this disorder over their lifespan.  

This project’s team has developed a mobile system-based prototype, Guess What, that uses a smartphone’s camera and fluidly engages with a child, facilitating prosocial learning and measuring the child’s developmental learning progress. The goal is to help the children be better at identifying social cues and behavior that will support them in forming new relationships. The team is constructing a novel Recurrent Neural Network models for identifying child’s emotions, head movements, hand movements, and eye gaze estimations. The project currently has active users in over 25 countries helping the team gather information from a diverse group of children of different races, cultures and ethnicities. These active users have so far played over 13,000 mobile games across 10 different game decks including faces/emojis, animal, sports, jobs, gestures and movement, objects, and code word decks. The medical experts on the team have also completed a clinical feasibility study on 72 children ages 3-8 and the results show a significant effect on social knowledge for children who played the game at least four times a week over one month period. 

“The project provides an innovative, engaging, and affordable intervention platform, driven by sophisticated machine learning algorithms, that helps family members and caregivers to work together with these children. These are very young children and working with family members is both fun and rewarding in terms of handling their (children’s) issues arising from this broad-spectrum disorder.”, noted NSF Program Director Prabha (Balakrishnan) Prabhakaran. 

Today, more than 750,000 Americans suffer from end-stage renal disease. For  most of these patients, treatment consists of biweekly dialysis, which severely affects their quality of life. Black Americans are 3 times as likely as white Americans to suffer from this disease. For the fortunate few, kidney transplants significantly improve quality of life, but the median wait time for a suitable donor kidney is more than 4 years. Additionally, many donated kidneys go unused because an acceptable match cannot be found.  

In some cases, personalized treatment may be offered to desensitize the patient to the donor kidney by a selective antigen-removing regimen. This project seeks to expand the pool of acceptable donor kidneys by building data-driven decision models to improve transplantation outcomes, thereby increasing the number of successful transplants, decreasing transplantation wait times, and lowering overall system costs. The project involves a collaboration between researchers at George Mason University, the University of Maryland, and the Virginia Commonwealth University Health Hume-Lee Transplant Center—a designated kidney paired donation center.   

“This work has the potential to significantly improve quality of life for patients suffering from end-stage kidney disease. Moreover, what the team will learn in designing algorithms for effective matching of kidneys with a desensitization regimen will provide insight into transplantation of other organs, such as the heart, lungs, and liver,” said NSF Program Director Georgia-Ann Klutke. 

Elizabeth Mynatt, Ph.D., is the newly appointed Dean of Khoury College of Computer Sciences at Northeastern University. Her career spans 23 years at the Georgia Institute of Technology (Georgia Tech), where she most recently served as Regents’ Professor in the College of Computing and executive director of the Institute of People and Technology. She has been recognized as an ACM Fellow, a member of the ACM SIGCHI Academy, and a Sloan and Kavli research fellow. She has published more than 100 scientific papers and chaired the CHI 2010 conference, the premier international conference in human-computer interaction. She currently serves as member of the National Academies Computer Science and Telecommunications Board (CSTB) and she chaired the recent update to the National Academies "Tire Tracks" report. Mynatt is also a member of the NSF CISE Advisory Committee. She is the past Chair of the Computing Community Consortium — an NSF-sponsored effort to engage the computing research community in envisioning more audacious research challenges.  

Mynatt is an internationally recognized expert in the areas of ubiquitous computing, personal health informatics, and assistive technologies. She has worked with several partners, including the Centers for Disease Control and Prevention, the Emory Brain Health Center, Parkview Health, and the Department of Biomedical Informatics at Columbia University to understand the design and adoption of socio-technical computing systems that enable people to better support their health and wellness needs. 

NSF support has been critical to Mynatt achieving her research ambitions and enabling her to evolve her research agenda. She received an NSF CAREER award that supported her research in fostering informal collaboration through Bayesian models of availability and likelihood of future interactions. A critical juncture in her career came with the NSF ITR funding of the Aware Home, launching decades of research into context-aware computing for home health and wellness needs. New NSF programs in Smart Health and Well-Being and Smart and Connected Health provided new avenues to explore the potential of mobile and wearable technologies and services. Mynatt’s research resulted in new approaches to foster sensemaking for newly diagnosed diabetics, fitness and healthy identities for adolescents, and personalized and adaptive engagement for breast cancer patients. This last project was jointly supported by the National Cancer Institute and was featured in a report to President Barack Obama by the President’s Cancer Panel. 

Mynatt is particularly excited to bring together these threads of her research in the newly funded NSF AI Institute for Collaborative Assistance and Responsive Interaction for Networked Groups (AI CARING). This institute “will develop a discipline focused on personalized, longitudinal, collaborative AI, enabling the development of AI systems that learn personalized models of user behavior, understand how people’s behavior changes over time, and integrate that knowledge to support people and AIs working together.” Akin to her decades of past research, Mynatt will help guide the innovation of new intelligent, interactive systems that are deeply rooted in health and wellness needs and that foster useable and useful human-centered experiences.