Development of a Visual Field Simulation Model of Longitudinal Point-Wise Sensitivity Changes From a Clinical Glaucoma Cohort

Abstract

PurposeTo develop a new visual field simulation model that can recreate real-world longitudinal results at a point-wise level from a clinical glaucoma cohort.MethodsA cohort of 367 glaucoma eyes from 265 participants seen over 10.1 ± 2.5 years were included to obtain estimates of “true” longitudinal visual field point-wise sensitivity and estimates of measurement variability. These two components were then combined to reconstruct visual field results in a manner that accounted for correlated measurement error. To determine how accurately the simulated results reflected the clinical cohort, longitudinal variability estimates of mean deviation (MD) were determined by calculating the SD of the residuals from linear regression models fitted to the MD values over time for each eye in the simulated and clinical cohorts. The new model was compared to a previous model that does not account for spatially correlated errors.ResultsThe SD of all the residuals for the clinical and simulated cohorts was 1.1 dB (95% confidence interval [CI]: 1.1–1.2 dB) and 1.1 dB (95% CI: 1.1–1.1 dB), respectively, whereas it was 0.4 dB (95% CI: 0.4–0.4 dB) using the previous simulation model that did not account for correlated errors.ConclusionsA new simulation model accounting for correlated measurement errors between visual field locations performed better than a previous model in estimating visual field variability in glaucoma.Translational RelevanceThis model can provide a powerful framework to better understand use of visual field testing in clinical practice and trials and to evaluate new methods for detecting progression.

Comparison of Visual Field Point-Wise Event-Based and Global Trend-Based Analysis for Detecting Glaucomatous Progression

“…Instead, we recently developed a computer simulation model that allows “real-world” visual field results from glaucoma eyes to be reconstructed at a point-wise level over time. 11 We showed that the simulated visual field results closely reflect the variability characteristics of those seen in the longitudinal clinical cohort it was modelled from. This model thus provides a robust framework for evaluating the sensitivity of the point-wise event- and global trend–based analyses at matched specificities, because longitudinal visual fields that are truly stable can be simulated.…”

“…Evaluating these methods using computer simulations is particularly advantageous, because it allows their specificities to be evaluated in a scenario that reflects “real-world” longitudinal follow-up, instead of evaluating it in a short-term test-retest scenario. We have described the details of this simulation model in a previous publication, 11 but describe it briefly below.…”

“…The probabilities at each location for a test thus provided a template of patient performance, accounting for the correlation between the measured values at each location during a visit (i.e., a global visit effect, such as from varying levels of attention between visits); this was termed a “noise template.” A “noise template” could then be combined with a “sensitivity template” to simulate “real-world” visual field results, as described in detail previously. 11 …”

Comparing 10-2 and 24-2 Visual Fields for Detecting Progressive Central Visual Loss in Glaucoma Eyes with Early Central Abnormalities

“…27 However, this is unlikely to be problematic because of our use of the global measure of MD, and since both the longitudinal cohort evaluated and computer simulations were primarily performed within a short timeframe (under 5 years), where the assumption of linearity for MD is likely to be sufficient. Furthermore, we did not model pointwise visual field sensitivity values in this study (as we had done in a previous study 28 ) since more visual field data would be needed to develop a sufficiently robust model of pointwise sensitivities. However, we have also recently demonstrated that point-wise eventbased analyses of the visual field data performs similarly to global trend-based analyses, 29 and given that the latter analysis was performed for both the 10-2 and C24-2 study, it is unlikely that the conclusions reached in this study would differ significantly if point-wise visual field progression analyses were performed.…”

Estimating Rates of Progression and Predicting Future Visual Fields in Glaucoma Using a Deep Variational Autoencoder

“…Another extension to the method presented in this manuscript is to generate synthetic glaucomatous visual fields to be used as a benchmark dataset to validate new methods for glaucoma research. Existing methods attempt to generate longitudinal visual fields by modelling PW relationships across time with potentially non-linear fits 34,35 , however they typically ignore spatial correlations in the visual field or can only predict stable fields. The decoder from the VAE can be used to simulate glaucomatous visual field series that accounts for spatial dependencies and the highly-nonlinear PW trends.…”