Document Type : Original Article

Authors

• Ari Taha Guron ¹
• Mardin Abdullah Anwer ²
• Sazan Kamal Sulaiman ³
• Sami Jalal AbdulSamad ⁴

¹ Department of IT& Computer Science, college of IT&CS, Catholic University, Erbil, Kurdistan region, Iraq.

² Department of Software and Informatics, College of Engineering, University of Salahaddin, Erbil, Kurdistan region, Iraq.

³ Department of computer engineering, Colleges of Engineering, Knowledge University, Kurdistan region, Iraq.

⁴ Department of Eye Surgery, Hawler Eye Teaching Hospital, Erbil, Kurdistan region, Iraq.

Abstract

This study aims to create a machine learning-based method for categorizing ocular impairment. Congenital, refractive error, age, diabetes, and unknown are the five primary causes that specialists consider. The suggested technique automatically classifies patients into one of the five groups based on their unique features by evaluating the ODIR dataset of patient records, which includes numerous demographic and clinical information, and utilizing machine learning algorithms. Most previous studies in this area have focused on classifying illnesses; hence, this study's main contribution is its innovative focus on categorizing the causes of eye disorders.
To the best of our knowledge, no ocular dataset has a label that specifies the cause of eye disease. The classes of eye disease have been added by Ophthalmologists.
Better patient outcomes and more effective use of healthcare resources can be achieved by increasing the precision of physicians' diagnoses and streamlining their decision-making. Compared to the other classification methods, the Quadratic SVM model has the highest accuracy of 71.3%.

Keywords

• Eye Impairment
• Congenital
• Refractive Error
• Ageing
• Diabetes
• Machine Learning
• Clinical Variables

Main Subjects

• Engineering