Filtering data from the collaborative initial glaucoma treatment study for improved identification of glaucoma progression

Abstract

BackgroundOpen-angle glaucoma (OAG) is a prevalent, degenerate ocular disease which can lead to blindness without proper clinical management. The tests used to assess disease progression are susceptible to process and measurement noise. The aim of this study was to develop a methodology which accounts for the inherent noise in the data and improve significant disease progression identification.MethodsLongitudinal observations from the Collaborative Initial Glaucoma Treatment Study (CIGTS) were used to parameterize and validate a Kalman filter model and logistic regression function. The Kalman filter estimates the true value of biomarkers associated with OAG and forecasts future values of these variables. We develop two logistic regression models via generalized estimating equations (GEE) for calculating the probability of experiencing significant OAG progression: one model based on the raw measurements from CIGTS and another model based on the Kalman filter estimates of the CIGTS data. Receiver operating characteristic (ROC) curves and associated area under the ROC curve (AUC) estimates are calculated using cross-fold validation.ResultsThe logistic regression model developed using Kalman filter estimates as data input achieves higher sensitivity and specificity than the model developed using raw measurements. The mean AUC for the Kalman filter-based model is 0.961 while the mean AUC for the raw measurements model is 0.889. Hence, using the probability function generated via Kalman filter estimates and GEE for logistic regression, we are able to more accurately classify patients and instances as experiencing significant OAG progression.ConclusionA Kalman filter approach for estimating the true value of OAG biomarkers resulted in data input which improved the accuracy of a logistic regression classification model compared to a model using raw measurements as input. This methodology accounts for process and measurement noise to enable improved discrimination between progression and nonprogression in chronic diseases.

Deep learning in ophthalmology: The technical and clinical considerations

“…Researchers at University of Michigan used longitudinal data on IOP and VFs to accurately forecast VF progression for participants in the Collaborative Initial Glaucoma Treatment Study. 81 Using a similar approach on a clinical based sample of Japanese normal tension glaucoma patients, KF was better able to predict 2-year MD forecast than linear regression of MD. 82…”

Dynamic Forecasting and Control Algorithms of Glaucoma Progression for Clinician Decision Support

“…with a full description of the model and approach given in Schell et al (2013). The AUC (obtained from the Mann-Whitney U statistic) for the ProP function applied to the testing set was 0.919, which is clinically considered to be very good.…”

“…Although IOP and its derivatives were not used as a factor in the ProP function (Equation (15)), it was found to be important in the Kalman filter modeling of test measurement evolution because IOP interacts with VF and PSD. A thorough treatment of the key factors involved in glaucoma progression can be found in Musch et al (2009) and Schell et al (2013).…”