OptiDON-Guard: An Ensemble Model for Identification of Dysthyroid Optic Neuropathy Using Optical Coherence Tomography...

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• Dysthyroid optic neuropathy (DON) is a vision-threatening complication of thyroid eye disease.
• Accurate and early identification of DON is crucial for timely intervention.

Purpose of the Study:

• To evaluate the efficacy of machine learning models utilizing optical coherence tomography angiography (OCTA) for the diagnosis of DON.
• To develop and validate a robust diagnostic tool for DON.

Main Methods:

• A retrospective, dual-center study included 346 patients with thyroid eye disease.
• OCTA scans were analyzed using a two-tiered stacking ensemble model (OptiDON-Guard) with feature selection and pseudo-labeling.
• Model performance was assessed using AUC, accuracy, sensitivity, specificity, and precision on internal and external validation sets.

Main Results:

• The OptiDON-Guard ensemble model outperformed individual models.
• Internal validation: AUC of 0.872, accuracy 0.943, specificity 0.979.
• External validation: AUC of 0.854, accuracy 0.816, specificity 0.842.

Conclusions:

• OptiDON-Guard is a reliable, interpretable, and clinically applicable tool for DON identification.
• The model enhances diagnostic precision for DON, offering significant translational relevance.