Abstract
Abstract: :
Purpose: Glaucoma is worldwide one of the leading causes of blindness. Previous studies could demonstrate changes in the autoantibody repertoires in glaucoma patients. The aim of this study was to analyze the use of autoantibody repertoires for the diagnosis of glaucoma by means of artificial neural network algorithms. Methods: 139 patients were divided into four groups: healthy volunteers without any ocular disorders (n=41), patients with primary open angle glaucoma (POAG, n=38), ocular hypertension (OHT, n=22), and normal tension glaucoma (NTG, n=38). All groups were matched for age and gender. The sera of patients were tested against Western blots of retinal and optic nerve antigens. The autoantibody patterns were digitized and subsequently analyzed by multivariate statistical techniques and artificial neural networks. Results: All groups revealed complex autoantibody patterns against ocular antigens. After randomly dividing patients into test and training sets, the analysis by means of artificial neural networks was performed. The diagnostic power of this antibody approach for the diagnosis of glaucoma could be assessed by calculating receiver operating (ROC) curves. The artificial neural network could reach a sensitivity of 90%, a specificity of 85%, and an area under curve (r-value, ROC-curve) of 0.85. Conclusions: In this study, we could demonstrate that pattern matching algorithms such as artificial neural networks could be used to detect glaucoma based on autoantibody patterns specific for this disease. Thus, the use of autoantibodies and advanced pattern matching algorithms could be a beneficial approach for the diagnosis of glaucoma.
Keywords: autoimmune disease • clinical laboratory testing