ABSTRACT
With the U.S. Food and Drug Administration's (FDA) approval and widespread use of COVID mRNA vaccines (principally the Pfizer-BioNTech and Moderna COVID vaccines), mRNA techniques have become widely recognized by both the media and the general public. Correspondingly, many topics related to these techniques have attracted significant interest in many research areas, including microRNA (miRNA), which regulates many mRNA types. Although miRNA has been researched since early 2000, the studies focused on miRNA in the context of individual diseases are all very recent. What constitutes an appropriate miRNA pair for a biomarker to support disease diagnosis is still an open question in many biochemical and medical investigations, for example, Alzheimer's disease. Sometimes, synthetic (artificial) miRNA is used as a normalizer (denominator of biomarker). Sometimes, a ubiquitous normalizer with a robust concentration value across many pathologies is chosen. In the biomedical field, researchers have selected markers in different ways, often without rigorous mathematical or statistical study. In this paper, instead of using these pathology-insensitive miRNAs as normalizers, we propose a new miRNA-pairs-selection algorithm with a multivariate statistics approach to search for a pair or of pathology-sensitive miRNAs for a given pathology. We demonstrate the performance of this algorithm through a published experiment using published Mild Cognitive Impairment (MCI) data.