When it comes to the health care industry, artificial intelligence that uses machine learning is a game-changing proposition that can make diagnoses better and faster while bringing down health care costs and accelerating research.

At the same time, there are ethical concerns around allowing machines to make high-stakes decisions normally made by doctors. There is apprehension around the transparency of decisions and how they will impact patients.

It’s a balancing act that Aidong Zhang, the Thomas M. Linville Professor of Computer Science at the University of Virginia School of Engineering and Applied Science, takes seriously. Zhang, who holds joint appointments in UVA’s Department of Biomedical Engineering and School of Data Science, has spent her career working to make artificial intelligence more trustworthy in the health care arena.

Zhang_Aidong_Media.jpg

Aidong Zhang, the Thomas M. Linville Professor of Computer Science, holds joint appointments in biomedical engineering and the School of Data Science.
 

Zhang obtained her Ph.D. in 1994 from Purdue University and was a SUNY Distinguished Professor at the State University of New York at Buffalo before coming to UVA in 2019. She specializes in using machine learning, a type of artificial intelligence focused on the use of statistical algorithms to make informed decisions that imitate human behavior. Her work helps to inform health care research and decisions using machine learning tools such as data mining, the process of sorting through large data sets to identify patterns, and bioinformatics, computer science applications that analyze RNA and amino acid sequencing.

“I have a long history working with biological data. Now, using machine learning, we can integrate what we know about the biological factors and social determinants of health into a single understandable health graph that can be used by doctors and researchers alike,” Zhang said. “Using machine learning tools we can see the relationship between disease and different factors.”

Zhang is renowned for her pioneering work in bioinformatics and computational biology, which was cited in her 2021 election into the College of Fellows for the American Institute for Medical and Biological Engineering. The author of two books in the field, in 2011 she established the Special Interest Group on Bioinformatics, Computational Biology and Biomedical Informatics in the Association for Computing Machinery, where she is also a fellow.

Zhang recently received a host of awards including two National Institutes of Health and two National Science Foundation grants totaling nearly $3.8 million — all with the goal of making machine learning more transparent and applying these techniques to biomedical data.

“Bioinformatics is revolutionizing medicine in terms of enabling diagnoses and discovering treatments,” said Sandhya Dwarkadas, the Walter N. Munster Professor and chair of computer science at UVA. “As an early and prolific researcher in the field, Aidong has been a force in achieving these gains and with these new grants, she continues to make significant contributions in this important area.”

The Relationship Between Disease Risk Factors and Social Determinants of Health

Disease risk is impacted by a host of factors including genetics, diet, age, sex and lifestyle. But that’s just part of the healthcare puzzle. Social determinants of health, or circumstances outside of health that play a role in disease management, are also important. These factors may include access to healthcare, the neighborhood in which you live and economic viability.

Zhang’s goal is to integrate biological health information with social determinants of health to better understand disease risk factors.

“Health care information, for example, genetics, age and diet, could be used in research and in the diagnosis of disease to pinpoint what groups are most vulnerable,” she said.

For example, HIV has been shown to disproportionately impact certain groups, including those living in poverty and those with less access to testing and treatment. By understanding which patients are most at risk for the disease, the health care industry can properly direct resources.

Machine learning and data mining are also being applied to conditions such as Alzheimer’s disease, which is typically diagnosed in later stages in African American and Hispanic patients.

$3.8 Million From the NIH and NSF for Machine Learning Research to Improve Healthcare

Zhang’s recent awards include an NIH grant to build an adaptive machine learning platform that streamlines healthcare information. This project will develop a new web-based tool that automatically and continually updates and curates biomedical literature for healthcare researchers within a large archive.

An NSF grant will be used to develop machine learning that combines biological healthcare data and social determinants of health data into a program that helps doctors and researchers take an aerial view of a patient’s risk for certain diseases.

With another NSF award, Zhang and her team will establish a platform for machine learning that takes information gathered from single-cell RNA sequencing technologies to come up with explainable and understandable health predictions.

“Single-cell RNA sequencing has become popular because it measures the individual cells, but often the information generated is difficult to interpret,” Zhang said.

The algorithms produced through the platform will help predict genes and pathways at the single-cell level in different organisms that impact disease forecasting.

Zhang also received a pilot award from the NIH to use electronic health records to detect patients with undiagnosed Alzheimer’s disease and dementia, which are often missed until the disease has progressed. This technology would use machine learning to flag the records of these patients so that doctors can follow up.

Zhang’s research could help fill a treatment gap faced by minorities in the U.S. According to the Alzheimer’s Association, African Americans are about twice as likely as whites to have Alzheimer’s but only about a third more likely to be diagnosed for the disease. A similar gap exists for Hispanics.

Putting machine learning data to work for patients is what Zhang gets excited about when she wakes up in the morning. She has always loved working around technology, but in the end, it’s not machines that matter to her the most.

“The technology is important,” she said, “but only if it has a positive impact on people.”