DATA HAS BECOME an invaluable tool in the health care industry, enabling doctors to make informed decisions to better help their patients. But with this rise in data availability comes a parallel need for ways to manage and interpret the information.
Health informatics is a continuously evolving discipline at the intersection of data science, health information technology, health information management, and data analytics. It is focused on improving health and health care by bringing theory into practice through enabling technologies.
Several faculty members in the Computer Science and Engineering (CSE) and Industrial, Manufacturing, and Systems Engineering (IMSE) departments have worked extensively on research related to this field, including National Academy of Medicine member Marion J. Ball, M.D. Their work aims to increase quality of life for patients, enable doctors to make better decisions, and improve the overall efficiency of the health care industry.
A Leader in Health Informatics
Dr. Ball was named the Raj and Indra Nooyi Endowed Distinguished Chair in Bioengineering in August 2021. The Nooyi Chair was established through a gift from alumnus Raj K. Nooyi (’78 M.S., Industrial Engineering) and his wife, Indra Nooyi.
Ball, who is also a Presidential Distinguished Professor, is known for her decades of transformational research in the areas of health care and informatics. She joined UTA in 2020 and is the executive director of the Multi-Interprofessional Center for Health Informatics, which she established after her arrival. The center prepares future health care leaders through multi-interprofessional education, research, and practice and strives to enhance the health and well-being of communities through enabling technologies and health informatics.
Ball was the first American and first woman to be elected president of the International Medical Informatics Association, and the Health Information Management Systems Society named her one of the 50 most influential IT professionals in the last half-century. She has received numerous other honors, including the International Medical Informatics Association Lifetime Achievement Award and the American Medical Informatics Association’s Morris F. Collen Lifetime Achievement Award.
Prior to coming to UTA, she was the senior adviser for health care informatics at the Center for Computational Health at IBM Research.
“Dr. Ball is an outstanding scholar whose research is making a major contribution to the field of health care informatics,” says Nooyi. “Indra and I are honored to support her plans to advance UTA’s thought leadership in this important and emerging field.”
Applying Data to Decisions
Although much of the data used in health informatics is patient- based—at least, from a systems and process engineering viewpoint— informatics is more focused on trying to make better decisions when applied to public health. In other words, it’s not just concerned with modeling.
IMSE Assistant Professor Yuan Zhou is an expert in modeling who has done extensive work using agent-based simulation to mimic how viruses spread in a population. Her research previously focused on the flu and its spread in locations like shopping centers and college campuses, but she soon found that her model could be generalized for any context and scaled up and down as necessary.
“This modeling is pretty powerful in expressing data at a very granular level,” Dr. Zhou says. “It can be used to help identify significant contributing factors to the spread of viruses that can be transmitted from person to person, such as flu, Ebola, and COVID-19, and evaluate the effectiveness of different mitigation strategies.”
Zhou’s experience came in handy when the coronavirus began to spread throughout the United States. Working with IMSE Professors Victoria Chen and Jay Rosenberger, she used the framework from her flu research to create an optimized model to identify strategies to help communities safely reopen and mitigate transmission of the virus. The researchers identified key contacts—such as people who were high risk and had to work, or people who were protecting others who were isolating—and extracted aggregate information from Yuan’s agent-based model.
With this, they formulated and solved a linear programming (LP) optimization to allocate personal protective equipment, testing, and vaccines to various types of key contacts engaged in different levels of activity.
The LP optimization takes only 30 seconds to solve and can be used to study many hypothetical scenarios in aggregate over time and for any region. With it, the researchers were able to model how to effectively and safely bring businesses and schools back to in-person activities without sacrificing a community’s overall health.
“The decision aspect of how we use this data is important to consider. It’s not just prediction for prediction’s sake; instead, we’re discovering how to apply the data and evaluate decisions. Our approach allows us to think dynamically because situations change,” Dr. Chen says. “The continuously evolving nature of the pandemic can render static decisions out-of-date, so it is important to be able to quickly re-evaluate alternate decisions and then clearly message these to the public.”
Managing Brain Data
Like the organ itself, data collected about the brain is incredibly complex and requires sophisticated methods to sort and analyze. These involve topology—a mathematical model that gives a picture of the data—or machine learning, which is a statistical model showing trends. But neither can handle the amount of data available with brain images.
Brain research relies on functional magnetic resonance imaging data that is highly complex, multiscale, and heterogeneous. This includes pathology or radiology images, along with omics data, such as genomics, proteomics, or metabolomics, captured from the same patient.
This data is at such a high resolution - an image might measure 1 million by 1 million pixels (for comparison, a cellphone screen measures 1,000 by 1,000 pixels)—that each piece of data may be 1 terabyte or more. Combining several of these large files for a holistic view creates massive amounts of data too large for current technology to handle.
Junzhou Huang, an associate professor in the CSE Department, hopes that combining topology, which analyzes 3D data and provides global information, and machine learning, which relies on historical data, will allow him to change the current model and predict the outcome of future data.
“Topology and machine learning are very different methods, but together they could help break down the data into more manageable units,” he says. “Topology could allow doctors or researchers to quickly identify the data they need, and then machine learning could fill in the fine details, making the process faster and more accurate.”
Building a Knowledge Base
Fillia Makedon, a CSE professor, is the co-founder of the Center for Assistive Technologies to Enhance Human Performance, or iPerform. Its research focuses on increasing, maintaining, or improving the functional capabilities of people with disabilities; enhancing the productivity of well-bodied people; and stimulating research to improve quality of life and prepare welltrained employees of the future.
She and CSE Professor Vassilis Athitsos received a National Science Foundation grant to use artificial intelligence to help experts assess learning difficulties in children very early in their lives. At the heart of the project—which also involves faculty from Yale University—is a computer vision recognition and machine-learning system that assesses children while they’re performing certain physical and computer-game-like exercises. The researchers collect and analyze the activity and interaction data to recognize patterns of inattention, hyperactivity, or impulsive actions, features common to executive function disorders like ADHD. By monitoring and analyzing how children behave during such gamelike exercises, the team hopes to build a knowledge base for health care professionals that enables them to apply predictive methods and make recommendations for effective intervention while also helping clinicians target the therapy.
The research has yielded positive results, and the team is beginning to expand the scope to young adult and adult subjects.
“Health informatics is a great application of basic computational methods,” Dr. Makedon says. “The power of databases and machine learning to predict, monitor, and provide targeted interventions allows us to automate how we collect and analyze data.”
iPerform is a National Science Foundation-funded Industry-University Cooperative Research Center that involves collaborations between industry and faculty at UTA, UT Dallas, and several other universities. Makedon is also director of the Heracleia Human-Centered Computing Laboratory, which works with researchers at universities around the U.S. on projects to improve quality of life and performance of special-needs persons and those around them.
In addition to these projects, Makedon’s research has focused on industrial settings. Through vocational simulation experiments, she studied how best to train and prepare workers, both with and without disabilities, for the industry of the future, where a worker must safely and efficiently collaborate with advanced robots.
The experiments were designed to assess in real time the cognitive skills—including attention and task awareness—and the physical and collaborative abilities of a person to determine their ability to work with a robot, to identify their weaknesses, and to recognize any special needs for personalized training and/or rehabilitation.
“No matter how talented a clinician is, each one has a different emphasis and viewpoint, and they aren’t with patients 24 hours a day,” Makedon says. “Computers can unravel certain hidden patterns that a superb clinician could miss, and with sensors it’s easy to unobtrusively collect and analyze information that can allow much more targeted and effective treatments.”