Artificial intelligence-powered renal pathology for glomerular disease classification: a systematic review and meta-analysis.

What was studied

This was a review and pooled analysis of AI tools for kidney pathology. The tools read biopsy images to sort glomerular diseases. The team asked how well AI works for each type. They also compared AI with pathologists. They searched PubMed, Embase, Web of Science and the Cochrane Library through 15 July 2026. 15 studies were included, with 39,536 validation sample units.

What they found

AI did well for membranous nephropathy (MN), IgA nephropathy (IgAN) and minimal change disease (MCD). Each had at least 10 validation datasets. For MN, sensitivity was 0.96 and AUC was 0.98. For IgAN, sensitivity was 0.92 and AUC was 0.96. For MCD, sensitivity was 0.92 and AUC was 1.00. AI also scored higher accuracy than senior pathologists for these types.

What it means for patients

The authors said pathology-based AI shows strong potential for sorting kidney disease types. They said current head-to-head evidence cannot show AI beats pathologists. This is most true for senior pathologists. They said future studies across several centers are needed. They should add clinical data of many types and standard outside checks.

Why it is not certain yet

Comparator evidence was sparse and should be read with care. Most studies were retrospective. Results varied a lot across datasets, imaging types, model designs and validation methods. The sample units were not necessarily unique patients.

Source

Artificial intelligence-powered renal pathology for glomerular disease classification: a systematic review and meta-analysis.

We restate facts from the source record in our own words. The original wording belongs to its publisher.