Retinal biomarkers in schizophrenia spectrum disorder — structural, functional, and computational insights

Schizophrenia spectrum disorder (SSD) is a complex neurodevelopmental disorder characterised by hallucinations, social withdrawal, and deficits in working memory and decision-making. Its underlying causes are not yet fully understood, but mounting evidence suggests that genetic, immunological, and environmental influences during the neurodevelopmental period contribute to its pathogenesis. Identifying measurable biomarkers that reflect the disorder's clinical features and progression is crucial for understanding its etiopathogenesis. As part of the central nervous system, the retina may serve as an important structure for identifying such biomarkers, given the structural and functional abnormalities observed in individuals with SSD. The unique properties of the human eye—translucence and low myelin and glial cell content—allow non-invasive examination of central nervous system microstructure and function. Growing evidence highlights retinal abnormalities in SSD, reinforcing their potential as candidate biomarkers and providing insights into the disorder's neurodevelopmental and neurodegenerative biology. Recent advances in artificial intelligence and machine learning have demonstrated remarkable capabilities in detecting these retinal changes, with convolutional neural networks achieving up to 95% classification accuracy and AUC values of 0.98 in controlled research settings. In this review, we discuss the retinal pathology associated with SSD, explore how these findings could offer new perspectives for future research into the molecular pathology of the disorder, examine the role of artificial intelligence and machine learning in retinal biomarker analysis, and explain their potential implications as candidate diagnostic biomarkers warranting further validation.