What question did this study set out to answer?

The aim is to develop predictive models to identify risk factors for increased intraocular pressure after silicone oil tamponade in retinal detachment patients.

March 16, 2026Open Access

A clinical prediction tool for ocular hypertension following silicone oil tamponade in rhegmatogenous retinal detachment

HJHuan JuThe Affiliated Yongchuan Hospital of Chongqing Medical University XQXuefeng QinThe Affiliated Yongchuan Hospital of Chongqing Medical University XWXing WangJiangsu University

Key Points

The aim is to develop predictive models to identify risk factors for increased intraocular pressure after silicone oil tamponade in retinal detachment patients.
Retrospective analysis of 533 patients with rhegmatogenous retinal detachment
IOP measured at multiple time points post-surgery
Construction of machine learning models and a nomogram using clinical data
Evaluation of model performance using AUC (Area Under Curve) metrics
57.8% of patients experienced elevated intraocular pressure after surgery
81.5% of IOP elevations occurred within two weeks post-operation
The Stepwise GLM + SVM model had an AUC of 0.761, outperforming the nomogram at 0.729
Key predictors of IOP elevation included axial length, age, white-to-white corneal diameter, PVR, and central corneal thickness

Abstract

This study aimed to develop predictive models for identifying risk factors of postoperative IOP elevation in patients with rhegmatogenous retinal detachment (RRD) following silicone oil (SO) tamponade. This retrospective study involved 533 RRD patients treated at the First Affiliated Hospital of Chongqing Medical University between September 1, 2024, and June 19, 2025. IOP was measured postoperatively on day 1, week 1, and months 1, 3, 6, and before SO removal using the non-contact tonometer Corvis Scheimpflug Technology (Corvis ST, OCULUS, Wetzlar, Germany) to acquire biomechanically corrected IOP (bIOP). Multiple machine learning (ML) models and one nomogram were constructed using demographic, clinical, and surgical factors. Postoperatively, 57.8% of patients developed elevated intraocular pressure (IOP), with 81.5% occurring within two weeks. A Stepwise GLM forward + SVM model based on clinical features outperformed the Nomogram (AUC 0.761 vs. 0.729). The top five predictors were axial length (AL), age, white-to-white corneal diameter (WTW), proliferative vitreoretinopathy (PVR), and central corneal thickness (CCT). In RRD patients receiving silicone oil tamponade, AL, age, WTW, PVR, and CCT were closely associated with postoperative IOP elevation. The ML models outperformed nomogram in accuracy and clinical interpretability, offering a reliable tool utilizing bIOP for early IOP risk assessment and postoperative management in RRD patients. Their integration of anatomical and biometric features supports real-world clinical deployment and individualized care planning.

Demander à l'IA

Bookmark

View Full Paper