We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
The Relevance of Cataract as a Risk Factor for Age-Related Macular Degeneration: A Machine Learning Approach.
- Authors
Martínez-Velasco, Antonieta; Martínez-Villaseñor, Lourdes; Miralles-Pechuán, Luis; Perez-Ortiz, Andric C.; Zenteno, Juan C.; Estrada-Mena, Francisco Javier
- Abstract
Age-related macular degeneration (AMD) is the leading cause of visual dysfunction and irreversible blindness in developed countries and a rising cause in underdeveloped countries. There is a current debate on whether or not cataracts are significant risk factors for AMD development. In particular, research regarding this association is so far inconclusive. For this reason, we aimed to employ here a machine-learning approach to analyze the relevance and importance of cataracts as a risk factor for AMD in a large cohort of Hispanics from Mexico. We conducted a nested case control study of 119 cataract cases and 137 healthy unmatched controls focusing on clinical data from electronic medical records. Additionally, we studied two single nucleotide polymorphisms in the CFH gene previously associated with the disease in various populations as positive control for our method. We next determined the most relevant variables and found the bivariate association between cataracts and AMD. Later, we used supervised machine-learning methods to replicate these findings without bias. To improve the interpretability, we detected the five most relevant features and displayed them using a bar graph and a rule-based tree. Our findings suggest that bilateral cataracts are not a significant risk factor for AMD development among Hispanics from Mexico.
- Subjects
MEXICO; RETINAL degeneration; MACHINE learning; CATARACT; DEVELOPING countries; ELECTRONIC health records; INTRAOCULAR lenses; SINGLE nucleotide polymorphisms
- Publication
Applied Sciences (2076-3417), 2019, Vol 9, Issue 24, p5550
- ISSN
2076-3417
- Publication type
Article
- DOI
10.3390/app9245550