Building an AI Model as a Tool

techy spiral graphic

The Department of Ophthalmology is building an artificial intelligence (AI) model (Digital Twin Eye) to function in real-time as a tool for early detection, progression monitoring, and decision support.

The project’s overall goal is to develop a framework for an automated personalized medicine system, allowing for the most effective treatment based on individual health data paired with predictive analytics. The Digital Twin model will leverage decades of clinical history and medical records to build a dataset scalable in size (number of patients), fields (different data points) to train, and test AI algorithms on temporal characteristics.

Over time, the trained AI will provide an individualized treatment course. This comprehensive patient data set will include treatment history, demographics, genetics, medical history, lifestyle, sex, socio-economic standing, and more. Importantly, social determinants of health will be incorporated in the model as these have a strong influence on informed decision-making, compliance, and monitoring. Accuracy of personalized prediction models using the Digital Twin approach is key to increasing the speed of diagnosis and care with high reliability.

The Quality team in the Department of Ophthalmology includes Dr. José-Alain Sahel, Distinguished Professor and Chair, Brian Rudolph, Executive Administrator, Dr. Vishal Jhanji, Vice Chair, Quality Improvement, and Laxmi Velankar, Ophthalmology Improvement Specialist. Other physicians and scientists from the Department are working on various aspects of the Digital Twin project: Sandeep Bollepalli, Jay Chabblani, Kunal Dansingani, Kiran Vupparaboina, and Andrew Williams.

The team has assessed the current state by measuring successful first surgeries and identifying variabilities in processes associated with cataract surgery and intravitreal injection procedures. Ongoing projects include measuring outcomes of cataract surgeries, reviewing and standardizing care protocols associated with ophthalmic surgeries, and reducing variabilities in processes associated with cataract surgery procedures and intravitreal injection procedures.

Next steps include conducting a thorough assessment of clinical and surgical protocols, with the goal of standardizing. Creating new processes and ensuring that all physicians take part in adopting them is another goal.

“The Digital Twin program is a very ambitious integrated project aiming at improving access, assisting in diagnostic and prognostic assessments, decision making (including social determinants of health), and continuous monitoring of quality,” Dr. Sahel said. “The support we obtained is of paramount importance in pursuing these important goals.”

This program is supported by the Eye & Ear Foundation, the Jewish Healthcare Foundation, the Pittsburgh Foundation for the quality project, the Shear Family Foundation for the predictive model, and the Hillman Foundation for the access to care frame.