Deep learning detection of retinitis pigmentosa inheritance forms through synthetic data expansion of a rare disease dataset.

Classification of inheritance patterns is important for clinical characterization and genetic counseling in inherited retinal diseases (IRDs). In practice, inheritance assessment integrates pedigree information, clinical evaluation, and genetic testing. However, a definitive molecular diagnosis is not achievable in a subset of patients, even with contemporary sequencing approaches, and family history may be incomplete or ambiguous. These limitations motivate investigation of complementary phenotype-based approaches that may provide additional contextual information, while not replacing molecular diagnosis when available. Deep learning (DL) applied to fundus imaging presents a promising approach for automated inference of inheritance modes, as recent advances in oculomics have demonstrated applications of DL in uncovering subtle phenotypic patterns associated with retinal conditions. However, development has been hindered by the low prevalence of IRDs and the scarcity of annotated datasets in individual clinical settings. In this study, we focus on retinitis pigmentosa (RP), a highly heterogeneous disorder in both clinical presentation and genetic etiology. We present a first-in-class deep learning approach that leverages Vision Transformer (ViT) models to distinguish autosomal from X-linked RP using color fundus photography. To overcome challenges posed by limited data, we introduce an innovative variational autoencoder–based data expansion strategy, which improves inheritance pattern classification based on color fundus photos from 0.67 AUC to 0.79 AUC. Our findings demonstrate the potential of deep learning to uncover subtle phenotypic differences linked to genetic inheritance, complementing existing genetic testing approaches, and introduce a novel training data augmentation method to render deep learning accessible to rare diseases.