Development of machine learning models to predict posterior capsule rupture based on the EUREQUO registry

The authors evaluated the performance of three probabilistic classifiers, including a Bayesian network, logistic regression model, and multi-layer perceptron network, to predict posterior capsule rupture (PCR) before cataract surgery. The classifiers were trained on 2,853,376 surgeries reported to the European Registry of Quality Outcomes for Cataract and Refractive Surgery between 2008 and 2018. The multi-layer perceptron network performed the best, followed by the Bayesian network and the logistic regression model. Direct risk factors for PCR were also identified, including preoperative best-corrected visual acuity, year of surgery, operation type, anesthesia, target refraction, other ocular comorbidities, white cataract, and corneal opacities. The study suggests that implementing the multi-layer perceptron network in clinical practice could decrease the PCR rate.