Overview
- The peer-reviewed study in Aging Cell, led by Duke Health with the University of Minnesota, analyzed blood from more than 1,200 community-dwelling adults aged 71 and older.
- Researchers used causal AI and machine learning to assess 828 small RNAs alongside 187 clinical variables, with survival outcomes verified through national mortality records.
- A six‑piRNA panel alone predicted two‑year survival with accuracy up to about 86%, and a combined model achieved cross‑validated AUC 0.92 in discovery and 0.87 in external validation.
- Lower circulating levels of specific piRNAs consistently associated with longer 2‑ and 5‑year survival, nine piRNAs were flagged as potential targets, and four were identified by causal analyses as likely direct determinants.
- piRNA measures outperformed familiar indicators such as age and cholesterol for short‑term risk, showed weaker value at 10 years, and will be the focus of follow‑up work on modulation by lifestyle or medications including GLP‑1 therapies.