April 2014
Volume 55, Issue 13
Free
ARVO Annual Meeting Abstract  |   April 2014
Artificial Neural Network Approach for Differentiating Open-Angle Glaucoma
Author Affiliations & Notes
  • Yoo Kyung Song
    Department of Ophthalmology, Yonsei Univ College of Medicine, Seoul, Republic of Korea
  • Samin Hong
    Department of Ophthalmology, Yonsei Univ College of Medicine, Seoul, Republic of Korea
  • Ein Oh
    Department of Ophthalmology, Yonsei Univ College of Medicine, Seoul, Republic of Korea
  • Tae Keun Yoo
    Department of Ophthalmology, Yonsei Univ College of Medicine, Seoul, Republic of Korea
  • Gong Je Seong
    Department of Ophthalmology, Yonsei Univ College of Medicine, Seoul, Republic of Korea
  • Footnotes
    Commercial Relationships Yoo Kyung Song, None; Samin Hong, None; Ein Oh, None; Tae Keun Yoo, None; Gong Je Seong, None
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science April 2014, Vol.55, 4305. doi:
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Yoo Kyung Song, Samin Hong, Ein Oh, Tae Keun Yoo, Gong Je Seong; Artificial Neural Network Approach for Differentiating Open-Angle Glaucoma. Invest. Ophthalmol. Vis. Sci. 2014;55(13):4305.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose: Visual field test for open-angle glaucoma (OAG) and glaucoma suspect (GS) has been a major workload of hospital eye services. The objective of this study was to select patients who should receive periodic visual field test due to clinically significant optic nerve damage in order to increase the effectiveness of treating OAG. To achieve the best performance in differentiating OAG from GS without visual field test, we used an artificial neural network (ANN).

Methods: We investigated Fifth Korean National Health and Nutrition Examination Survey for OAG prediction models. In this cross-sectional study, 386 participants, who underwent visual field test using frequency doubling technology, were included in the study population. For risk prediction model development, the association between clinical features and OAG was examined by multivariate logistic regression (LR) and ANN.

Results: Among 386 participants from the study population, 94 subjects had OAG. The predictors selected by LR included sex, age, menopause, duration of hypertension, myopic spherical equivalent, intraocular pressure, vertical cup-disc ratio, temporal superior RNFL defect, and temporal inferior RNFL defect. The ANN model was the best discriminator between OAG and glaucoma suspect with an AUC of 0.890. This model predicted OAG with an accuracy of 84.0%, a sensitivity of 78.3%, and a specificity of 85.9%.

Conclusions: To our knowledge, this is the first study to develop the mathematical models for OAG risk prediction among patients with suspected glaucoma using population-based health records. ANN might be cost-effective screening tools identifying OAG among patients with suspected glaucoma. The machine learning technique using ANN can contribute to the advancement of clinical decision-making tools with a good discriminative ability for OAG.

Keywords: 464 clinical (human) or epidemiologic studies: risk factor assessment  
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×