Ocular Hypertension or Glaucoma Risk Factor Prediction Models for Patients with Fuchs Uveitis Syndrome.

Abstract

Purpose: Fuchs uveitis syndrome (FUS) is a chronic inflammatory disorder that can lead to ocular hypertension (OHT)/glaucoma. We aimed to investigate the risk factors for OHT/glaucoma and develop prediction models for these complications in patients with FUS.

Methods: Patients with FUS being referred to the uveitis centre at the First Affiliated Hospital of Chongqing Medical University between April 2008 and March 2023 were enrolled and divided into training and test sets in a 7:3 ratio. Cox regression analysis was used to construct a traditional clinical prediction model (TCPM). Multiple artificial intelligence algorithms were applied to construct 101 machine learning (ML) models. The Goldmann-Witmer coefficient (GWC) was used to investigate the relationship between antiviral antibodies in the aqueous humour and OHT/glaucoma in these patients.

Results: The study included 1099 patients with FUS. Fifteen clinical features were chosen for the construction of TCPM and 101 ML models. Univariate Cox regression analysis identified seven features correlated with OHT/glaucoma. Multivariate Cox regression analysis revealed that three parameters were significantly correlated with OHT/glaucoma. The TCPM demonstrated a satisfactory predictive performance. The least absolute shrinkage and selection operator+random survival forest model was the best-performing ML model. Antiviral antibodies in the aqueous humour (GWC-positive) showed no statistically significant association with the development of OHT/glaucoma.

Conclusion: The most significant risk factors for OHT/glaucoma in patients with FUS were ocularor systemic corticosteroid treatment, recurrence at least three times per year, and male sex. The TCPM demonstrated the highest utility and performance for predicting OHT/glaucoma in these patients.