Ecological Informatics · May 3, 2026
Immune-derived biological age as a machine learning predictor of health in the honey bee model
By Frunze O, Kim H, Yun Y et al.
Biological age, which reflects physiological rather than chronological time, provides a sensitive measure of health and aging in honey bees. We developed a predictive model based on the expression of four immune-related genes ( Dome, Relish, SOD1, Apid1 ), normalized to a housekeeping gene and 0-day-old honey bees. Trained on healthy, short-lived bees (0–25 days post-emergence), the model remained sensitive in detecting elevated biological age in diseased individuals and identifying aging patterns in long-lived, overwintered bees. All processing steps—including Ct normalization, model training, and prediction in day-equivalent—were implemented in a reproducible R-based script with a Shiny interface, generating visual plots and downloadable Excel outputs. This molecular tool enables stress detection, health monitoring, and intervention assessment across diverse ecological and pathological contexts.