Integrated blood miRNA-mRNA profiling in long COVID (LC) patients vs. recovered controls, using hierarchical clustering to define molecular subclusters and a random forest classifier to distinguish them.
Clustering revealed an immune-hematopoietic subcluster (LC1) with greater symptom burden, functional impairment, lower quality of life, elevated fibrin D-dimer and thrombin time, and lower plasma sodium vs. LC2. A 9-miRNA + LRRFIP2 random forest classifier achieved AUROC 0.91 for identifying LC1.
Sample size is not reported in the abstract; potential confounding by age, sex, comorbidities, and medication use was evaluated but not fully eliminated. Generalizability is uncertain given single-cohort design and cross-sectional blood sampling.
Long COVID is not one disease — a biologically distinct immune-hematopoietic subtype (LC1) with measurable coagulation and electrolyte changes may be identifiable via blood biomarkers. Clinicians should consider subtype heterogeneity when assessing prognosis and symptom burden in long COVID patients.
