A digital twin for pregnancy: AI test-drives medicines in a virtual immune system — and finds two familiar drugs that may prevent early labor

About one in ten pregnancies worldwide ends in early labor, and doctors have had no new medicine to prevent it for decades. A Stanford Medicine team tried a different route: they measured responses from a quarter of a billion immune cells, built a map of the immune system in pregnancy accurate down to single cells, and used it to create 'digital twins' — virtual copies of an individual pregnant woman's immune system — so drug combinations can be test-driven safely on a computer before anyone swallows a pill. Screening nine candidate combinations this way pointed to two familiar medicines used together: low-dose aspirin and lansoprazole, a common reflux drug. In the virtual system the pair's effect on delaying labor was 15 times stronger than aspirin alone, and in an animal test it did delay labor. The finding was strong enough that the team has won FDA approval to start a clinical trial in real patients. The study, led by Brice Gaudilliere and published in Science Advances, builds on the idea of an immune 'clock' of pregnancy: when that clock runs too fast, labor starts too soon.

SDG 3 · Good health and well-being — Mechanism: tech innovation