Peer Review #1 of "Improving clinical refractive results of cataract surgery by machine learning (v0.1)"

Abstract

To evaluate potential of Support Vector Machine Regression model (SVM-RM) and Multilayer Neural Network Ensemble model (MLNN-EM) as an intraocular lens (IOL) power calculation improvement for clinical workflow. Background. Current IOL power calculation methods offer limited accuracy and especially in eyes with an unusual ocular dimension the accuracy may decrease. In case of an improperly calculated power of the IOL in cataract or refractive lens replacement surgery, there is a risk of re-operation or further refractive correction. This may potentially induce complications and discomfort to the patient. Methods. A dataset containing information about 2194 eyes was obtained using data mining process from Electronic Health Record (EHR) system database of Gemini Eye Clinic. The dataset was optimized and split into Selection set (used in design of models and training), and Verification set (used in the evaluation). Set of mean prediction errors and distribution of predicted refractive error were evaluated for both models and Clinical results. Results. Both models performed significantly better for the most evaluated parameters compared to the clinical results. There was no significant difference between both evaluated models. In the ± 0.50 D prediction error category both SVM-RM and MLNN-EM were slightly better than Barrett Universal II formula (which is often presented as the most accurate calculation formula). Conclusion. In comparison to the method currently used in a clinical setting, both SVM-RM and MLNN-EM has achieved significantly better results in IOL calculations and therefore there is a strong potential to improve clinical cataract refractive outcomes.