Jump ARCHES Funding Provides an Excellent Opportunity to Merge CS Research with Real-World Applications

For two Illinois Computer Science professors – Gang Wang and Yuxiong Wang – newly funded projects through the Jump ARCHES research and development program offer up a powerful opportunity for real-world applications through collaborative research.

At the beginning of the year, Jump ARCHES announced more than $1.4 million in funding for 20 research projects.

Housed within the Health Care Engineering Systems Center at The Grainger College of Engineering, each Jump ARCHES project features a partnership between OSF HealthCare, the University of Illinois Urbana-Champaign and its College of Medicine in Peoria.

“We're among the lucky group of researchers who are in a very supportive college that is encouraging interdisciplinary work,” Gang Wang said. “We work hard to form a trusting team by spending time and resources together to solve a problem that is bigger than the one we each have individually. It’s incredibly helpful to have a formal mechanism for this collaboration through Grainger Engineering.”

Each of the two Illinois CS professors involved this year have uniquely positioned their research to make an impact.

High Trust Patient Outreach
Gang Wang, UIUC; Jonathan Handler, OSF HealthCare; Nick Heuermann, OSF HealthCare; Cody Zevnik, OSF HealthCare

Gang Wang’s research at Illinois CS centers on Security and Privacy, Internet Measurement, and Data Mining. His work takes a data-driven approach to addressing emerging security threats in massive communication systems (social media, email services), crowdsourcing systems, mobile applications, and enterprise networks.

These capabilities in CS appealed to collaborators at OSF HealthCare, who, according to Gang, face consistent hurdles in forming secure patient outreach – as many other healthcare providers.

Specifically, this project “aims to survey, design, and potentially prototype feasible solutions to enable secure patient outreach for patients across all levels of socioeconomic status. We also want to provide patients and doctors with a list of best practices to use the solution to communicate securely.”

“Of course, we’re aware that there is a ton of sensitive data involved in the healthcare system,” Gang said. “Knowing that there is a sense of urgency toward making digital outreach more approachable and secure, I’m excited to navigate this design space to find a new way that can work for the patients impacted by this conflict.

“It’s also become clear that one of the specific challenges is that many of the patients from low socioeconomic status do not have access to smartphones or reliable laptops.”

Rather than jumping right into building an outreach tool, this group will first analyze the patient population at OSF HealthCare to see what kind of technological capabilities are realistic. Then the group will plan around these results to construct a solution that is more tailored to the actual patient capabilities.

Gang said this workgroup works well together, because of how well his research matches up with the priorities of his collaborators – even mentioning that he and Jonathan Handler from OSF crossed paths through past experiences with Microsoft Research.

“This group is set up for success as a team, because we have people ingrained in proper digital infrastructures,” Gang said. “My collaborators from OSF HealthCare will build a great data analytics pipeline to build and maintain a system designed to operate well for the users.”

Early Detection and Prediction of Facial Expression for Parkinsonism Powered by Few-Shot Learning
Yuxiong Wang, UIUC; Christopher Zallek, OSF HealthCare; George Heintz, UIUC; Manuel Hernandez, UIUC 

Yuxiong Wang admits that a phrase people have become most familiar with when considering his research focus of Artificial Intelligence and Computer Vision is “big data.”

The problem is that for most, "big data" doesn't exist - it is much more limited. And that’s where Yuxiong's research comes in. He specializes in few-shot learning and predictive learning with limited annotated data.

Considering the limitations in the healthcare setting when it comes to collecting and using data – primarily, security limitations – this project will look to address a significant neurological disorder in Parkinson’s disease.

The goal of this project revolves around an “AI supported system that tracks facial expressions of neurological patients and reports findings to the neurologists. In this project we focus on discriminating facial expressions associated with Parkinsonism. Our long-term vision is that the check-in process point-of-care camera becomes a channel to provide additional documentation of baseline neuro exam findings.”

“Recently, there has been exciting progress in introducing data-driven, machine learning techniques for healthcare problems. However, currently established approaches often rely on large amounts of annotated images or videos, which is often infeasible for practical clinical application,” Yuxiong said. “Our research will try to close this gap by introducing a computational tool for discriminating between normal versus abnormal facial expressions in relation to Parkinson’s disease with a minimal size of annotated data. We believe that data efficient AI is urgently needed for healthcare and its patients, and that our project is an important step to establish and democratize few-shot learning in this area.”

Based on the visual cues these images offer healthcare providers, early detection of the disease will be more likely. And early detection, according to Yuxiong, can make a big difference in early diagnosis and the overall care path of the patient.

He also sees other application opportunities based on the model this project provides.

“A broader vision for this work could be something that impacts a number of research areas on our campus at the University of Illinois,” Yuxiong said. “It’ll be exciting to see what builds through this project involved with Jump ARCHES and OSF Healthcare, because it will shed light on a previously invisible solution and will lead to small-data AI empowered healthcare research.”