AI-Designed Drug Shifts Six Biological-Age Clocks

A small lung-disease trial links rentosertib to younger proteomic readings, but cannot yet distinguish treatment effects from slower aging.

Six clocks move in the same direction

An experimental drug discovered and designed with artificial intelligence produced younger biological-age estimates across six independently developed proteomic clocks, according to a study published in Nature Biotechnology. Rentosertib, developed by Insilico Medicine, inhibits TNIK and is being tested primarily as a treatment for idiopathic pulmonary fibrosis, an incurable disease that progressively scars the lungs.

Researchers applied the clocks to longitudinal blood-protein data from 42 participants in a randomized Phase 2a trial. After 12 weeks, every clock detected shifts consistent with lower predicted biological age among treated patients. The strongest reported readings suggested reductions of several years, while placebo recipients generally showed little improvement. The analysis also found dose-related changes in forced vital capacity, a standard measure of lung function.

The result does not establish that rentosertib slows aging. All participants had pulmonary fibrosis, so improvements in inflammation, lung function or disease burden could have made their blood profiles appear younger without changing the underlying aging process. The cohort was small, the clocks measure biomarkers rather than lifespan or healthspan, and the drug has not been tested as a longevity treatment in healthy people. The authors explicitly say those effects cannot be disentangled without further trials.

Why it matters

The study nevertheless advances two parts of AI-enabled medicine beyond laboratory prediction. Rentosertib combines an AI-identified target with an AI-designed molecule that has reached human testing, while the deposited proteomic data and analysis pipelines offer a reproducible method for adding aging-related endpoints to conventional drug trials. If larger studies validate the approach, developers could evaluate disease treatment and possible geroprotective effects simultaneously. For now, the durable milestone is methodological: an AI-originated drug has generated measurable human biomarker data, not evidence that artificial intelligence has produced an anti-aging medicine.

← All briefings