miRBiT: a rules-based single-sample serum miRNA classifier for pan-cancer detection with multi-cohort validation

The Research group of Prof. Himanshu Kumar, Department of Biological Sciences, developed “miRBiT,” a rules-based single-sample serum microRNA (miRNA) classifier for pan-cancer detection with multi-cohort validation. This method helps in early cancer screening by utilizing circulating serum miRNAs as biomarkers leading to cost-effective, minimally invasive screening. Unlike conventional machine learning models that requires cohort-level normalization, miRBiT employs a robust rules-based framework capable of classifying individual patient samples independently via set of rules. The classifier was developed using GEO datasets of cancer and healthy serum samples containing miRNA expressions and validated across multiple independent cohorts representing diverse cancer types and populations. High diagnostic accuracy, sensitivity, specificity, and reproducibility, were achieved by this method, highlighting its potential for reliable pan-cancer detection in real-world clinical settings. Due to using single-sample level interpretation, the model minimizes batch effects and improves interpretation through biologically meaningful rules, paving way for clinical translation in future. miRBiT has the potential to support early cancer detection, improve patient outcomes, and facilitate precision medicine by enabling rapid, non-invasive, and single-sample cancer screening. This work marks a significant advancement in translational cancer diagnostics by integrating machine learning with liquid biopsy for biomarker-based screening providing a scalable framework for future clinical settings in oncology. For more details, kindly visit https://academic.oup.com/bib/article/27/2/bbag189/8662209