Fig. 2: Ability of SpotChecks, performed at home, to predict which eyes were subsequently listed for surgery. | Eye

Fig. 2: Ability of SpotChecks, performed at home, to predict which eyes were subsequently listed for surgery.

From: Pushing the envelope: the feasibility of using a mailed contrast sensitivity test to prioritise cataract waiting lists

Fig. 2

A Mean [±95% confidence intervals] SpotChecks scores for eyes that were/were not subsequently listed for surgery, plus associated t-test values (see also Fig S2 for further analysis). B Receiver Operating Characteristics [ROCs] showing the ability of SpotChecks to predict which eyes were/were not subsequently listed for surgery. Shaded regions indicate the 95% confidence intervals. The black line indicates when SpotChecks data alone were used. The blue dashed line indicates when a Support Vector Machine was used to combine SpotChecks data and Snellen acuity scores (model trained and evaluated using leave-one-out analysis, using the following Matlab functions: fitcsvm.m, fitSVMPosterior.m, kfoldPredict.m, perfcurve.m). Numerical values show the Area Under the ROC [AUROC], plus 95% confidence intervals. C Histograms showing the distributions of raw SpotChecks scores.

Back to article page