Predicting Visual Outcomes for Retinal Vein Occlusion Patients After Anti-VEGF Treatment

196 patient records with clinical and optical coherence tomography features for predicting poor functional visual outcomes after anti-VEGF treatment for macular edema secondary to retinal vein occlusion. The dataset was created by Haiyue Yu and published on figshare in May 2026. It includes a temporal validation split with training data from 2021-2024 and an independent test set from 2025.

Use Cases

• Developing predictive models for poor visual outcomes based on baseline BCVA, age, and RVO subtype.
• Analyzing non-linear relationships between quantitative OCT biomarkers like DRIL length and prognosis.
• Comparing performance of machine learning models like XGBoost and random forest on clinical prediction tasks.
• Applying SHAP analysis for model interpretability in clinical decision support systems.

Strengths

• Dataset includes 196 patient records with a rigorous temporal validation design.
• Model development utilized 15 candidate variables and identified five key predictors.
• The optimal XGBoost model achieved an AUC of 0.898 in the independent test set.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect temporal bias inherent to the single-source, retrospective study.

Provenance

Source

figshare

Collection Method

Retrospective clinical study.

Time Range

Patient cases from 2021 to 2025.

Freshness

Last updated 2026-05-18 04:17:03; freshness should be verified.