Screening for non-invasive rsRNA biomarkers to assess embryo quality utilizing ultra-sensitive pandora sequencing combined with machine learning.

Purpose: Accurate embryo selection is vital for the success of in vitro fertilization (IVF); however, existing morphological scoring methods are inherently subjective and fail to capture underlying molecular alterations. This study aimed to identify non-invasive molecular markers for embryo quality assessment by analyzing highly modified ribosomal small RNAs (rsRNAs) in embryo culture medium using ultra-sensitive sequencing and machine learning.

Methods: Ultra-sensitive Pandora sequencing was employed to profile rsRNAs in embryo culture medium. Machine learning algorithms were used to identify rsRNA biomarkers linked to embryo quality. Candidate rsRNAs were further validated via quantitative reverse transcription polymerase chain reaction (qRT-PCR).

Results: Four candidate rsRNAs (5S, 5.8S, 28-1S, 28-2S) were significantly associated with embryo quality through machine learning analysis, achieving high predictive accuracy (AUC = 0.955) in cross-validation. qRT-PCR confirmed that 5.8S and 28-2S rsRNA levels were markedly elevated in the culture medium of high-quality embryos.

Conclusion: Specific rsRNAs, particularly 5.8S and 28-2S, may serve as non-invasive biomarkers for embryo selection in IVF. These findings highlight the potential roles of rsRNAs in embryonic development and provide a molecular basis for improving IVF outcomes.