AI-Derived Retinal Microvascular Biomarkers for Chronic Kidney Disease

322 participants' retinal fundus photographs analyzed using an AI platform to quantify microvascular parameters associated with chronic kidney disease. The dataset includes 110 healthy controls, 142 patients with CKD stages 1–2, and 70 patients with CKD stages 3–5. It was created by Qiumei Gu and last updated on May 18, 2026.

Use Cases

• Building predictive models for CKD risk based on AI-derived retinal vascular parameters like arteriovenous ratio (AVR) and vascular density.
• Investigating the relationship between systemic microvascular changes and specific retinal features such as arterial tortuosity.
• Evaluating the discriminatory performance of retinal biomarkers for differentiating early and advanced CKD stages.
• Developing non-invasive screening tools for early-stage CKD detection using fundus photography.

Strengths

• The dataset is derived from a structured case-control study with 322 participants, providing a clear comparison group.
• Analysis includes specific, quantified retinal parameters (e.g., AVR, aTort, aVD) with statistical associations reported.
• The model performance is evaluated with a reported AUC of 0.776 for CKD discrimination.

Limitations

• The dataset is described in a 121.6 KB DOCX file, suggesting the underlying data tables or images are not directly accessible.
• Column-level documentation and sample data are unavailable, requiring manual inspection after download.
• The study is single-center, which may limit the generalizability of the findings.

Provenance

Source

figshare, author Qiumei Gu.

Collection Method

Fundus photographs from a single-center case-control study were analyzed using an AI-based platform.

Freshness

Last updated 2026-05-18 04:16:53; freshness should be verified.