Screening for Non-Invasive rsRNA Biomarkers to Assess Embryo Quality Using Ultra-Sensitive Pandora Sequencing Combined with Machine Learning

Accurate embryo selection is vital for successful in vitro fertilization (IVF), but current morphological scoring methods are somewhat subjective and do not reflect molecular changes. This study employs ultra-sensitive Pandora sequencing to detect highly modified rsRNAs in culture media, aiming to identify molecular markers for non-invasive embryo quality assessment. Machine learning identified four candidate rsRNAs (5S, 5.8S, 28-1S, 28-2S) associated with embryo quality, with cross-validation demonstrating high predictive accuracy (AUC = 0.955). Quantitative RT-PCR further confirmed that 5.8S and 28-2S levels were significantly higher in the culture media of high-quality embryos. These findings suggest that specific rsRNAs could serve as non-invasive markers for embryo selection, offering new insights into rsRNA functions in embryo development. Overall design: This is a case-control study. In this study, the high-quality embryos and their culture medium were set as the case group while the low-quality embryo and their culture medium were set as the normal control group.