Identification of diagnostic biomarkers in diabetic retinopathy using machine learning and RNA sequencing data

Diabetic Retinopathy (DR) constitutes a significant contributor to the impairment of visual acuity, yet its intricate development is not fully explored, making detection and treatment challenging. This study aims to identify new diagnostic biomarkers for DR by using differentially expressed genes (DEGs) in peripheral blood samples of DR patients and controls through RNA sequencing. A total of 498 DEGs were identified, showing significant involvement in protein binding, transcription regulation, apoptosis, and angiogenesis pathways. Different machine learning methods like Support Vector Machine (SVM), Logistic Regression, Neural Networks, Random Forest, and Gradient Boosting were used to build a reliable diagnostic model based on the top 7 DEGs chosen through the Elastic Net method. SVM performed the best, with an accuracy of 0.86 and an area under the curve (AUC) of 0.99 on the test dataset. Further analysis highlighted diagnostic genes DRAXIN, TBCEL, ZNF775, TMSB4XP4, SLIT1, FSD1L , and NRADDP strongly associated with DR. Moreover, studying immune cell infiltration revealed a notable connection between these genes and immune cells, especially DRAXIN , and the immune environment, implying their role in regulating immune responses in DR. The results indicate the potential of these genes as biomarkers for DR detection and provide insights into its development, opening the door for targeted treatment approaches.