Ametropia detection using a novel, compact wavefront autorefractor.

INTRODUCTION: Despite the well-known reproducibility issues of subjective refraction, most studies evaluating autorefractors compared differences between the device and subjective refraction. This work evaluated the performance of a novel handheld Hartmann-Shack-based autorefractor using an alternative protocol, which considered the inherent variability of subjective refraction. METHODS: Participants underwent an initial measurement with a desktop autorefractor, two subjective refractions (SR1 and SR2) and a final measurement with the QuickSee Free (QSFree) portable autorefractor. Autorefractor performance was evaluated by comparing the differences between the QSFree and each of the subjective refractions with the difference between the subjective refractions (SR1 vs. SR2) using Bland-Altman analysis and percentage of agreement. RESULTS: A total of 75 subjects (53 ± 14 years) were enrolled in the study. The average difference in the absolute spherical equivalent (M) between the QSFree and the SR1 and SR2 was ±0.24 and ±0.02 D, respectively, that is, very similar or smaller than the SR1 versus SR2 difference (±0.26 D). Average differences in astigmatic components were found to be negligible. The results demonstrate that differences between QSFree and both subjective refractions in J0 and J45 were within ±0.50 D for at least 96% of the measurements. The limits of agreement (LOAs) of the differences between QSFree and SR1, as well as QSFree and SR2, were higher than those observed between SR1 and SR2 for M, J0 and J45 . CONCLUSIONS: A protocol was designed and validated for the evaluation of a refractive device to account for the variability of subjective refraction. This protocol was used to evaluate a novel portable autorefractor and observed a smaller difference between the device and subjective refractions than the difference between the two subjective refraction measurements in terms of mean bias error, although the standard deviation was higher.

Prediction of manifest refraction using machine learning ensemble models on wavefront aberrometry data.

PURPOSE: To assess the performance of machine learning (ML) ensemble models for predicting patient subjective refraction (SR) using demographic factors, wavefront aberrometry data, and measurement quality related metrics taken with a low-cost portable autorefractor. METHODS: Four ensemble models were evaluated for predicting individual power vectors (M, J0, and J45) corresponding to the eyeglass prescription of each patient. Those models were random forest regressor (RF), gradient boosting regressor (GB), extreme gradient boosting regressor (XGB), and a custom assembly model (ASB) that averages the first three models. Algorithms were trained on a dataset of 1244 samples and the predictive power was evaluated with 518 unseen samples. Variables used for the prediction were age, gender, Zernike coefficients up to 5th order, and pupil related metrics provided by the autorefractor. Agreement with SR was measured using Bland-Altman analysis, overall prediction error, and percentage of agreement between the ML predictions and subjective refractions for different thresholds (0.25 D, 0.5 D). RESULTS: All models considerably outperformed the predictions from the autorefractor, while ASB obtained the best results. The accuracy of the predictions for each individual power vector component was substantially improved resulting in a ± 0.63 D, ±0.14D, and ±0.08 D reduction in the 95% limits of agreement of the error distribution for M, J0, and J45, respectively. The wavefront-aberrometry related variables had the biggest impact on the prediction, while demographic and measurement quality-related features showed a heterogeneous but consistent predictive value. CONCLUSIONS: These results suggest that ML is effective for improving precision in predicting patient's SR from objective measurements taken with a low-cost portable device.