Digital twins forecast individual immune trajectories

ImmuneTwins digital twins, built from single-cell profiles, cut immune-age prediction error by 32% and 42% in two external human populations versus the best existing method.

• Why it matters: Immune-response models often rely on one baseline measurement and pooled patterns, limiting their ability to forecast how an individual’s immune state evolves after new observations.
• What they did: The approach personalizes a shared virtual-cell model with each person’s incoming single-cell observations, then forecasts immune change across external cohorts and repeated influenza vaccination.
• The result: During repeated influenza vaccination, it predicted immune-cell change direction 35% more accurately and reduced error by 27% as observations accumulated, while myeloma twins anticipated patient-specific trajectories and localized B-cell remodeling in exceptional longevity.

AI summary of the abstract · may contain errors · not medical advice · read the paper.