Abstract
Purpose :
Open globe injuries (OGI) often have a significant impact on clinical outcomes, including visual acuity (VA), which remain challenging to predict upon initial presentation. We tested the hypothesis that machine learning (ML) could accurately predict visual acuity outcomes using presenting clinical features and demographics using a retrospective cohort analysis.
Methods :
Clinical data for patients suffering OGIs treated at Massachusetts Eye and Ear between 2012-2022 were collected and manually validated to ensure accuracy and completeness of the data. The following were selected as input variables to the ML model: age, gender, race, ethnicity, mechanism of injury, zone of injury, presenting VA (in LogMAR), afferent pupillary defect, intraocular foreign body, lensectomy, and time from presentation to operating room. The data were split into 80% training and 20% hold-out testing. The following ML regression models were evaluated with 10-fold cross validation: random forest, adaptive boosting, gradient boosting, extra trees, and linear regression. Performance was measured with mean absolute error (MAE) and mean squared error (MSE) with the optimal hyperparameters. Relative importance of input variables were evaluated for the best performing model.
Results :
A total of 903 patients were reviewed and included in analysis. Random forest regression showed the best performance from 10-fold cross validation of all models assessed (Table 1). Subsequent evaluation of the random forest model on the testing holdout showed an MAE and MSE of 0.58 LogMAR and 0.62 LogMAR respectively with an optimal hyperparameter of 500 decision trees. Relative feature importance revealed age, time to the operating room, and presenting VA as most predictive for final VA (Figure 1).
Conclusions :
Our results suggest that ML may be a useful tool for predicting VA based on presenting clinical features and demographics for patients with OGI. Random forest model showed the best performed of all ML models. Future work should work to prospectively validate ML VA predictions as well as improve predictive accuracy and generalizability through inclusion of additional patients, additional clinical metrics, and multi-institutional collaboration.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.