Related Experiment Videos
Modelling series of visual fields to detect progression in normal-tension glaucoma
A I McNaught1, D P Crabb, F W Fitzke
1Glaucoma Unit, Moorfields Eye Hospital, London, UK.
Summary
Statistical modeling can predict glaucoma progression. A simple linear model accurately forecasts visual field changes, aiding in early detection and management of this optic nerve disease.
Area of Science:
- Ophthalmology
- Statistical Modeling
- Medical Prognostics
Background:
- Glaucoma is a progressive optic nerve disease characterized by visual field loss.
- Accurate prediction of glaucoma's progression is crucial for timely intervention.
- Statistical modeling offers potential for describing and forecasting disease trajectory.
Purpose of the Study:
- To identify statistical models that accurately describe glaucomatous sensitivity decay.
- To evaluate the predictive capability of different models for future visual field status.
- To determine the optimal model for forecasting glaucoma progression.
Main Methods:
- Analysis of Humphrey visual fields from 12 untreated normal tension glaucoma patients (mean follow-up 5.7 years).
- Curve-fitting software applied 221 models to pointwise sensitivity data against time.
- Models assessed on data fit (R2) and predictive accuracy of future visual field status.
Main Results:
- Complex polynomial models provided the best data fit (median R2 0.93).
- Simpler models, particularly linear expressions, showed superior accuracy in predicting future deterioration.
- Linear models adequately fitted most data and yielded the most precise future predictions.
Conclusions:
- A linear model of visual sensitivity over time effectively detects and forecasts glaucoma progression.
- Linear modeling enables estimation of the clinically significant rate of visual sensitivity loss.
- This approach provides a framework for managing glaucoma progression.