Development of a recommendation system for selecting a formula in cataract surgery

Abstract

Accurate intraocular lens (IOL) power calculation remains a critical factor for achieving optimal refractive outcomes in cataract surgery. This study analyzes existing methods and software solutions for selecting formulas used to calculate IOL power. To solve this problem, a support medical decision-making recommendation system (SMDRS) was developed to analyze patient biometric data and predict the most suitable calculation formula. Among the evaluated machine learning approaches, the random forest (RF) algorithm demonstrated the highest stability and classification accuracy, leading to its selection as the core predictive engine. The system was validated using retrospective clinical data and evaluated in a functioning ophthalmology clinic. Performance evaluation demonstrated that the system increased the success rate of surgical outcomes in complex cases from 73.5% to 90.5%, thereby confirming its impact on improving the efficiency of optical calculations in clinical practice. By minimizing human error and standardizing decision-making, the proposed solution offers a robust tool for ensuring consistently superior surgical results.