Diagnostic accuracy of machine learning classifiers for cataracts: a systematic review and meta-analysis

ABSTRACT Objective The objective of this study was to systematically review and meta-analyze the diagnostic accuracy of current machine learning classifiers for pediatric and adult cataracts. Methods MEDLINE, EMBASE, CINAHL, and ProQuest Dissertations and Theses were searched systematically and thoroughly. Conferences held through Association for Research in Vision and Ophthalmology, American Academy of Ophthalmology, and Canadian Society of Ophthalmology were searched. Studies were screened using Covidence software and data on sensitivity, specificity and area under curve were extracted from the included studies. STATA 15.0 was used to conduct the meta-analysis. Results Our search strategy identified 150 records from databases and 35 records from gray literature. Total of 21 records were used for the qualitative analysis and 11 records (100 134 images) were used for the quantitative analysis. In adult patients with cataracts, the pooled estimate for sensitivity was 0.948 [95% CI: 0.815–0.987] and specificity was 0.960 [95% CI: 0.924–0.980] for cataract screening using machine learning classifiers. For pediatric cataracts, the pooled estimate for sensitivity was 0.882 [95% CI: 0.696–0.960] and specificity was 0.891 [95% CI: 0.807–0.942]. Conclusion The included studies show promising results for the diagnostic accuracy of the machine learning classifiers for cataracts and its potential implementation in clinical settings. Prospero registration CRD42020219316