Keratoconus Severity Staging Using Random Forest and Gradient Boosting Ensemble Techniques
Muhsin, Zahra J. Ghafir, Ibrahim Al Bdour, M. AlRyalat, S. AlShawabkeh, M. Al Taee, M.
Muhsin, Zahra J.
Ghafir, Ibrahim
Al Bdour, M.
AlRyalat, S.
AlShawabkeh, M.
Al Taee, M.
Publication Date
2025-05-20
End of Embargo
Supervisor
Rights
© 2025 IEEE. Reproduced in accordance with the publisher's self-archiving policy.
Peer-Reviewed
Yes
Open Access status
embargoedAccess
Accepted for publication
Institution
Department
Awarded
Embargo end date
2026-05-20
Additional title
Abstract
Accurate keratoconus (KC) staging is vital for improving patient care and guiding clinical decision making. Choosing the right Machine Learning (ML) algorithm is key to effectively tackling this challenge and ensuring optimal performance. This paper presents a detailed comparison of eight ML algorithms commonly used in KC detection, based on a clinical dataset collected by the authors over the past decade. The study investigates each algorithm's effectiveness in distinguishing KC severity stages. Results showed that ensemble learning algorithms outperformed others, with Random Forest (RF) achieving the highest accuracy at 98.82%, followed closely by Gradient Boosting (GB) at 98.24%. These models also had the highest classification quality scores, with RF at 0.985 and GB at 0.978. These findings underscore the strength and effectiveness of ensemble classifiers for KC severity staging. Furthermore, the top-performing model (RF) exceeded results from recent studies on KC severity, highlighting its potential for clinical application.
Version
Accepted manuscript
Citation
Muhsin ZJ, Qahwaji R, Ghafir I, et al (2025) Keratoconus Severity Staging Using Random Forest and Gradient Boosting Ensemble Techniques. IEEE 22nd International Multi-Conference on Systems, Signals & Devices (SSD), Monastir, Tunisia, 17-20 Feb 2025. pp. 593-598.
Link to publisher’s version
Link to published version
Link to Version of Record
Type
Conference paper
Qualification name
Notes
The full text will be available at the end of the publisher's embargo: 20th May 2026