With many clinicians hesitant to trust AI to help them make decisions about patient health, WVU research — led by Avishek Choudhury, assistant professor and NSF Faculty Early Career Development award winner — is identifying the moments when providers lose that faith in AI tools. (WVU Photo/Jennifer Shephard)
A West Virginia University researcher who studies health care providers’ trust in artificial intelligence has been named a Faculty Early Career Development award winner by the National Science Foundation.
Avishek Choudhury, an assistant professor in the Department of Industrial and Management Systems Engineering within the WVU Benjamin M. Statler College of Engineering and Mineral Resources, works at the intersection of patient safety, artificial intelligence and clinical decision-making.
The CAREER award is considered the NSF’s most prestigious honor supporting junior faculty. It will allow Choudhury to develop a machine-learning framework that can identify changing patterns of trust in AI as clinicians make decisions about patient care.
“Artificial intelligence is increasingly used to support decision-making in health care, especially in time-sensitive settings such as emergency care and intensive care units, but people do not always rely on these systems appropriately,” Choudhury said.
“Some users may place too much trust in incorrect recommendations. Others may ignore useful guidance. These mismatches can affect decision quality and patient safety. By improving how clinicians interpret and respond to artificial intelligence, my research supports safer and more reliable decisions.”
Choudhury said his goal is to humanize the algorithms behind AI. It’s not about persuading clinicians to place more trust in AI, he emphasized. Instead, his research focuses on helping users maintain an appropriate level of trust based on the quality of the recommendation and the clinical context.
“Trust should not be viewed as simply high or low,” he said. “The appropriate level of trust depends on the situation. Clinicians should be able to use AI as a valuable decision-support tool while continuing to apply their own expertise and professional judgment.”
To study how trust changes over time, Choudhury and his research team will conduct simulation-based experiments involving AI-assisted health care decisions. During these experiments, the team will collect behavioral and physiological data, such as eye-gaze patterns, heart-rate variability, brain activity and changes in the electrical characteristics of the skin associated with physiological arousal.
The researchers will combine these signals with survey responses and information about how users interact with AI recommendations. Machine-learning methods will then be used to identify early indicators of trust, overreliance and insufficient reliance.
“Trust is dynamic,” Choudhury said. “It can change from one decision to the next, based on factors such as the accuracy of previous AI recommendations, the user’s workload and the complexity of the clinical situation. If we can identify those changes early, we can design AI systems that respond more effectively to users’ needs.
One major factor driving clinicians’ lack of trust in AI, Choudhury added, is liability.
“If an AI system provides a recommendation and something goes wrong, the clinician remains accountable for the decision,” Choudhury said. “That is why trustworthy AI must do more than produce an answer. It must support clinicians in evaluating the recommendation and deciding whether it is appropriate for a particular patient and situation.”
Choudhury also cautioned that consistently accepting AI recommendations without sufficient reflection could weaken independent clinical reasoning over time.
“We want to develop AI systems that understand not only the clinical task but also the human being who is interacting with the technology,” he said, noting that medical decision-makers are right to regard AI with skepticism, even when the system is making trustworthy recommendations.
Choudhury said his work is about understanding what’s happening when providers lose faith in the technology, in order to create systems that better serve their needs and those of patients.
CAREER awardees have brought nearly $25 million to WVU since 1997. Overall, 53 WVU faculty have received the award.
“The dedication of our faculty to cutting-edge research and their brilliance never cease to amaze me,” said Interim Vice President for WVU Research Ming Lei. “Dr. Choudhury’s accomplishment is a testament to our standing as an R1 university. This award will not only propel his career, but also help improve health care for people.”
-WVU-
mm/6/15/26
MEDIA CONTACT: Rachel Brosky
Media Manager
WVU Strategic Communications and Marketing
304-293-3990; RaBrosky@mail.wvu.edu
Call 1-855-WVU-NEWS for the latest West Virginia University news and information from WVUToday.
© 2026 West Virginia University. West Virginia University is an Equal Opportunity Employer. Last updated on June 15, 2026.
Leave a Reply