For generations of ophthalmologists, learning cataract surgery has depended on a remarkably human process.
A resident sits at the operating microscope while an experienced surgeon watches closely— observing an instrument that drifts from the center of the eye, a movement that could be more precise or a surgical step that takes longer than expected. Afterward comes the feedback: what went well and what to try differently next time.
At the University of Iowa Department of Ophthalmology and Visual Sciences, that kind of close surgical mentorship has long been central to resident education.
Now, Iowa will help investigate whether artificial intelligence can add something new for ophthalmologists in training.
The department is participating in a National Institutes of Health-funded, multi-institutional study evaluating PhacoTrainer, an AI-assisted platform designed to analyze videos of cataract surgery and provide objective measurements of surgical performance.
The study is led by Sophia Wang, MD, MS, of Stanford University, with Oregon Health & Science University also participating. At Iowa, Jaclyn Haugsdal, MD, will serve as site principal investigator and recruit ophthalmology residents to participate.
The question researchers hope to answer: Can AI-generated feedback help residents become better surgeons?
Turning surgery into data
Becoming proficient at cataract surgery requires mastering a complex sequence of precise movements inside one of the body's smallest and most delicate spaces. Residents develop those skills through supervised surgery, simulation, repetition, and feedback from faculty.
PhacoTrainer adds another source of information.
Using computer vision and deep learning, the platform analyzes recorded cataract surgeries, identifying portions of the procedure, and tracking surgical instruments and structures within the eye. It can then turn aspects of surgical technique into data: How far did an instrument travel? How quickly did it move? How well centered was the eye or phacoemulsification probe? How long did individual steps take?
Over time, those measurements could help residents see not only that they are improving, but how they are improving.
Research from Wang and collaborators has already shown that the technology can detect differences in surgical performance. In a 2025 study, PhacoTrainer analyzed cataract surgery videos from 28 residents and 29 attending surgeons. AI-generated measurements identified differences between the groups in areas such as instrument movement and centration, and several measurements correlated with assessments made independently by experienced ophthalmologists.
The new study takes the next step: determining whether those measurements can be turned into useful feedback for residents.
A natural fit for Iowa
For Iowa, the project builds on an established emphasis on surgical education and simulation.
Haugsdal, a clinical associate professor of ophthalmology and visual sciences, has made resident education a focus of her work at Iowa. She has helped develop the department's ophthalmology wet-lab curriculum, procedure-specific training opportunities and a structured simulation curriculum for teaching ophthalmic laser procedures. In 2022, those efforts earned her a Carver College of Medicine Excellence in Clinical Coaching Award.
PhacoTrainer brings new technology to that same educational goal: helping residents better understand their performance and improve their surgical skills.
One potential advantage is the ability to follow those skills over time. Instead of simply knowing how many cataract procedures a resident has completed, educators could potentially see how individual aspects of technique change from one operation to the next. A resident might see that unnecessary instrument movement has decreased, for example, or that the eye remains more consistently centered during surgery.
The technology is meant to complement, not replace, the expertise of faculty surgeons. AI can measure an instrument's movement; an experienced teacher can explain why that movement occurred, whether it mattered and how to approach the next case differently.
Putting AI to the test
Iowa's participation will help researchers determine whether combining those forms of feedback improves surgical training in practice.
The multi-institutional nature of the study is important. Surgical teaching, equipment, techniques, and resident experiences can differ among training programs. Including residents from Iowa, Stanford and Oregon Health & Science University will help researchers evaluate PhacoTrainer in different real-world training environments.
The study also represents a different role for AI in medicine. While much attention has focused on using artificial intelligence to diagnose disease, interpret medical images or assist with clinical decisions, PhacoTrainer explores whether AI can provide data and feedback to help physicians develop the skills needed to perfect cataract surgery.
PhacoTrainer can track each operation and even the smallest movements, but experienced faculty are essential to put that data into context. Their judgment and guidance can help residents understand what the numbers mean and how to apply them in the operating room.
University of Iowa residents will now help determine just how useful that new set of eyes can be.