Two-parameter ridge estimation in seemingly unrelated regression models
COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION
Authors: Esfanjani, Robab Mehdizadeh; Najarzadeh, Dariush; Khamnei, Hossein Jabbari; Hormozinejad, Farshin; Talebi, Mahnaz
Abstract
Seemingly unrelated regression (SUR) models were applied when several linear regression equations were investigated at the same time. To reduce the multicollinearity influence in the SUR models, the one-parameter ridge (Ridge-1) solution was proposed and discussed by some researchers. As a generalization of the Ridge-1 solution, in the context of SUR models having multicollinearity problem, the two-parameter ridge (Ridge-2) solution was presented. Some simulations were performed to compare the proposed solution with the ordinary generalized least squares (GLS) and Ridge-1 solutions. Lastly, the proposed solution was applied on chronic renal failure effect data.
Optical Coherence Tomography Structural Abnormality Detection in Glaucoma Using Topographically Correspondent Rim and Retinal Nerve Fiber Layer Criteria
AMERICAN JOURNAL OF OPHTHALMOLOGY
Authors: Yang, Hongli; Luo, Haomin; Hardin, Christy; Wang, Yaxing; Jeoung, Jin Wook; Albert, Cindy; Vianna, Jayme R.; Sharpe, Glen P.; Reynaud, Juan; Demirel, Shaban; Mansberger, Steven L.; Fortune, Brad; Nicolela, Marcelo; Gardiner, Stuart K.; Chauhan, Balwantray C.; Burgoyne, Claude F.
Abstract
PURPOSE: This study evaluated the ability of topographically correspondent (TC) minimum rim width (MRW) and peripapillary retinal nerve fiber layer thickness (pRNFLT) criteria to detect optical coherence tomography (OCT) structural abnormality in glaucoma (GL) and glaucoma suspect (GLS) eyes. DESIGN: Retrospective cross-sectional study. METHODS: A total of 196 GL eyes, 150 GLS eyes, and 303 heathy eyes underwent pRNFL and 24 radial optic nerve head OCT imaging and manual correction of the internal limiting membrane, Bruch's membrane opening (BMO), and outer pRNFL segmentations. MRW and pRNFLT were quantified in 6 Garway-Heath or 12 30-degree (clock-hour) sectors. OCT abnormality for each parameter was defined to be less than the 5th percentile of the healthy eye distribution. OCT abnormality for individual eyes was defined using global, sectoral, and combined parameter criteria that achieved >= 95% specificity in the healthy eyes. TC combination criteria required the sectoral location of MRW and pRNFLT abnormality to be topographically aligned and included comMR (a previously reported TC combination consisting of MRW and pRNFLT parameter: [MRW + pRNFLT x (average MRW healthy eyes/average pRNFLT healthy eyes) MRW]. RESULTS: TC sectoral criteria (1 Garway-Heath MRW + corresponding Garway-Heath RNFLT), (one 30-degree MRW + any 1 corresponding or adjacent 30-degree pRNFLT), 30-degree and Garway-Heath comMR-TI and global comMR were the best performing criteria, demonstrating (96%-99% specificity), 86%-91% sensitivity for GL, 80%-84% sensitivity for early GL (MD >= -4.0 dB) and 93%-96% sensitivity for moderate-to-advanced GL (MD < - 4.0 dB). CONCLUSIONS: Clinically intuitive TC MRW and pRNFLT combination criteria identified the sectoral location of OCT abnormality in GL eyes with high diagnostic precision. (C) 2019 Elsevier Inc. All rights reserved.