aceA Risk Prediction Model for Normal-Tension Glaucoma Progression Integrating Genetic Markers and Corneal Biomechanics
Abstract: Objective Normal-tension glaucoma (NTG) is characterized by progressive optic nerve damage despite intraocular pressure remaining consistently within the normal range. Predicting disease progression in patients with confirmed NTG remains challenging. This study aimed to develop and validate an interpretable machine learning model integrating genetic risk scores and corneal biomechanical parameters to predict progression risk in patients with NTG, identify independent predictors, and quantify the contribution of…
Glaucoma by Osmosis. CC BY-SA 4.0 · https://creativecommons.org/licenses/by-sa/4.0
On August 24, 2026, a peer-reviewed study reported development and validation of an interpretable machine learning model to predict progression risk in 342 patients with normal-tension glaucoma enrolled at a tertiary hospital. The team integrated corneal biomechanical parameters, genetic risk scores, and tear molecular biomarkers, selecting predictors with LASSO and multivariable logistic regression, and compared random forest, SVM, and logistic regression models using AUC, calibration, and decision curve analysis with SHAP interpretation.
The work matters because NTG progresses despite normal intraocular pressure, making risk stratification difficult, and a quantitative, interpretable tool could guide personalized monitoring and treatment decisions. What remains uncertain is generalizability beyond a single tertiary center, performance in diverse populations and real-world workflows, and whether predicted risk actually changes outcomes, as the study reports potential clinical utility rather than prospective impact.
- Study enrolled 342 NTG patients at a tertiary hospital, split 238 training and 104 validation, to address prediction challenge where optic nerve damage progresses despite normal intraocular pressure.
- Candidate predictors selected via univariate analysis and LASSO regression with 10-fold cross-validation and the bb-1se criterion, then independent predictors identified with multivariable logistic regression.
- Three models random forest, support vector machine, and logistic regression were evaluated with AUC, calibration curves, decision curve analysis, and SHAP analysis for feature contribution interpretability.
An interpretable machine learning model integrating genetic risk scores, corneal biomechanical parameters, and tear molecular biomarkers was developed and validated to predict progression risk in patients with confirmed normal-tension glaucoma, providing a quantitative reference for risk stratification and personalised
The rundown
Researchers consecutively enrolled 342 NTG patients at a tertiary hospital, collecting baseline characteristics and six core indicators including corneal biomechanical parameters, genetic risk scores, and tear molecular biomarkers. After univariate screening and LASSO with 10-fold cross-validation, multivariable logistic regression identified independent predictors for model building.
Random forest, SVM, and logistic regression models were trained on 238 patients and tested on 104 patients, with performance assessed by AUC, calibration curves, and decision curve analysis, plus SHAP analysis to quantify individual feature contributions for clinical interpretability.
Sources
- Peer-reviewedSeminars in Ophthalmology2026-08-24
How should this claim be treated?
The debate