An artificial-intelligence powered algorithm will likely be capable of predicting the success of minimally-invasive glaucoma surgeries (MIGS) at multiple postoperative time points based on individual patient characteristics, according to a team of presenters at the American Society of Cataract and Refractive Surgery (ASCRS) 2026 meeting, held April 10 to 13, 2026 in Washington DC. The investigators report that their intentions are to construct the algorithm to aid in helping surgeons select the proper MIGS option to provide patients the best possible outcomes.

Researchers from the Centre Hospitalier de l’Université de Montréal (CHUM) and University of British Columbia (UBC) sought to identify key preoperative patient characteristics associated with MIGS success. Presenting author Alexander Cornea, MD, explained that the research team retrospectively collected data from 1000 patients who have undergone a MIGS procedure and had 1 to 5 years of follow-up visits.

The team reviewed the participants’ demographics, medical data (including the specific MIGS completed), surgery history, best-corrected visual acuities, intraocular pressure (IOP), and corneal thickness. They also examined the participants’ use of glaucoma medications, visual field metrics, and optical coherence tomography (OCT) images.

They then conducted a multiple linear regression analysis to determine which baseline variables had an effect on outcomes. 

Although the algorithm is still being developed, the group did share some preliminary findings. For patients who received an iStent (Glaukos; n=156), a multiple regression analysis with 6-month postoperative IOP as the outcome variable yielded an R squared of 0.429. Women and those with elevated baseline IOPs had higher postoperative IOPs at 6 months follow-up (β(female)=2.669; P =.0032; β(baseline IOP)=0.384; P =.0010). 

For patients who underwent gonioscopy-assisted transluminal trabeculotomy (n=19), the same analysis yielded an R squared of 0.986, and none of the baseline variables significantly affected postop IOP at 6 months follow-up.

The variables with the most weight in predicting a successful outcome include surgical technique used — GATT, Preserflo (Glaukos), and XEN (Allergan) were high-ranking predictors. Preoperative IOP and whether the procedure was performed alongside phacoemulsification were also significant variables.

This research highlights a shift toward “personalized medicine” in ophthalmology, with assistance from an AI algorithm aiding in predicting individual surgical outcomes.

“The algorithm showed promising predictive performance, achieving a 32% improvement over a dummy classifier at 1-year follow-up,” according to the presenters. 

The team noted that as more patients reach the 3-year and 4-year follow-up marks, the model’s long-term predictive validity will improve.