Multifaceted AI research positions WSU as a leader in new era of technological transformation

Every time Dr. Ming Dong, Wayne State University computer science professor, watches one of his artifical intelligence algorithms build a synthetic CT scan from a patient's MRI, he realizes he could be looking at the future of medical imaging.

The synthetic, AI-constructed scan shows all the meticulous detail of a traditional CT scan - bone density, tissue health, the condition of blood vessels - but with one glaring difference: The patient doesn't actually have to enter a CT machine.

Computer science professor Dr. Ming Dong is developing AI technology to create synthetic CT scan data that could reduce a patient's need for multiple scans and exposure to radiation.

"When you do radiation therapy, you need to do treatment planning, and to do treatment planning, a patient needs to go through CT scans," Dong explained. "But you can't just keep doing CT scans. CT scans are expensive, and they are just like X-rays. There is exposure to radiation. So, we synthesize the CT data from the MRI. The patient doesn't really need to go through the actual CT scan."

This breakthrough - now in clinical trials with partners at the University of Wisconsin - illustrates the power of Wayne State's expanding AI research ecosystem. Underscoring its standing as a premier research university, WSU has, in recent years, launched a host of cutting-edge AI research projects and initiatives. This includes the Institute for AI Data and Science (AIDaS), which was approved by the university's Board of Governors in October.

The AIDaS is intentionally interdisciplinary, connecting researchers from engineering, medicine, public health, education, business, the social sciences and the liberal arts.

Moreover, AIDaS is an integral part of one of the four key pillars that undergird WSU's new Grand Challenges initiative, which directs research toward pressing societal needs in critical areas of artificial intelligence, health, mobility and environmental sustainability.

"Artificial intelligence gives us the chance to reimagine health care and health education for Detroit and beyond," said Dr. Bernard J. Costello, WSU senior vice president for health affairs.

"Artificial intelligence gives us the chance to reimagine health care and health education for Detroit and beyond," said Dr. Bernard J. Costello, WSU senior vice president for health affairs.

"If we embed AI responsibly - with ethics, transparency and human judgment at the core - we can prepare the next generation of health professionals to lead in an AI-driven world, while ensuring every community benefits from safer, smarter and more accessible care."

As AI becomes ever more central to technological advancement as well as social evolution, Wayne State finds itself uniquely poised to lead the city, state and region into a new era of unprecedented transformation.

Trustworthiness and community care

Today, WSU's AI research is rooted in five major areas, as laid out by the Wayne Artificial Intelligence research initiative:

• Edge computing
• Mobility
• Health care
• Biomedical informatics
• Medical imaging
• Public safety

Together, these fields shape a campuswide vision of AI that is both cutting-edge and community-driven.

Ming Dong's work fits squarely in the center of the biomedical analytics pillar. Even though he recognizes how his group's synthetic CT technology could transform medical imaging, he remains vigilant about reducing the risks of generative AI. "A fake tumor is not something good," he said. "We need to eliminate hallucination in synthesized data."

While Dong's work in medical imaging focuses on enhancing what AI can show, WSU Associate Professor Dr. Alex Kotov uses his research to develop what the artificial intelligence program can say. Kotov is designing systems that speak directly to weight-loss patients, offering those struggling with obesity the kind of sustained, personalized support that clinicians don't always have time to provide.

Associate computer science professor Dr. Alexander Kotov is developing an AI program, dubbed NAOMI, that is designed to help users build readiness for weight loss.

His NIH-funded project, NAOMI - the Neural Agent for Obesity Motivational Interviewing - helps people build readiness for weight loss. "Boosting motivation is not easy," Kotov said. "Doctors are too busy. Nurses are too busy - that's where these agents have their shot."

Kotov envisions AI assistants that surgeons, EMTs or nurses can consult instantly, hands-free, during life-or-death moments: "If a surgeon is doing surgery and needs to ask something, they can talk to an agent and get reliable medical information instantly. If a medical worker in an ambulance has a case they've never seen before - if they don't know what to do, if the patient is dying - they need to immediately get credible medical advice and act upon it."

His work directly advances AIDaS's core area of intelligent agents in health care - AI designed to provide health behavior counseling and clinical decision support.

Meanwhile, Dr. Dongxiao Zhu, WSU professor, founding director of the Wayne AI research initiative and co-founder of AIDaS, focuses on the safety and reliability of AI amidst its frenzied growth. Large language models (LLMs) trained on the open internet absorb everything - insight, misinformation, harmful bias and copyrighted material.

Dr. Dongxiao Zhu (left), WSU computer science professor, stands alongside fellow co-director of AIDaS, Dr. Hengguang Li, chair of mathematics in the College of Liberal Arts and Sciences.

"Sooner or later, the whole internet will be flooded with synthetic text," Zhu warned. "That's very concerning."

One of his NSF-funded projects, Collaborative Research: III: Medium: Advancing Large Language Model Unlearning: Foundations and Applications, develops LLM unlearning, a cutting-edge technique allowing AI systems such as ChatGPT or Gemini to selectively "forget" toxic, private or copyrighted content without the expense of retraining massive models from scratch. "We teach the model to forget unwanted knowledge when needed," he said.

Zhu's research forms a cornerstone of AIDaS's mission: creating AI that is safe, ethical and in harmony with societal ethics. As Zhu is quick to point out, AI's value isn't just in its technological complexity but its ability for positive impact on communities such as Detroit.

"We are actually part of the city," Zhu said. "That allows us to make foundational breakthroughs based on what we have in Detroit."

Despite the complex technology at play, the objectives at the heart of WSU's AI research remain profoundly human, with their commitment to improved health care, more efficient transportation, enhanced privacy, smarter cities and reliable AI.

As Dong put it, "AI is here to help us do a more efficient, more effective job… as long as we keep our critical thinking."