A Stanford Medicine-led team has used AI models of the immune system during pregnancy to identify a drug combination for testing against premature labour. Aspirin plus lansoprazole emerged from a screen of nine regimens and subsequently delayed labour in an animal experiment. The researchers say they have obtained FDA authorisation to begin a human clinical trial.[1]
Birth is associated with a sequence of immune changes. Early pregnancy requires tolerance of a fetus whose biological characteristics differ from the mother's; later, inflammatory signals help prepare the body for labour. If inflammatory changes occur too early, they may contribute to premature delivery. A useful model must connect detailed cell responses to the question a clinician cares about: whether the pregnancy is progressing toward labour faster than expected.[1]
Connecting cell responses with the pregnancy clock
The system, called SIMCO, draws on measurements of responses from a quarter of a billion immune cells exposed to different drugs. The team combined that information with longitudinal data from 63 pregnancies whose outcomes were known. One layer models how cells respond to treatment; another connects patterns of immune activity with progression towards labour.[1]
With an initial blood sample, the model aims to estimate an individual’s position on that immune clock and simulate how a treatment might shift it. Its patient-specific predictions come from measured immune responses and clinical timelines. That focus gives researchers a way to compare treated and untreated scenarios for the same modelled pregnancy.[1]
Among the nine regimens, aspirin plus lansoprazole produced a predicted delay effect 15 times that of aspirin alone. This is a comparison within the immune-clock model, rather than a measured reduction in human preterm births. The model also identified immune pathways through which the combination could support pregnancy and inhibit processes associated with labour.[1]
Taking the prediction into biology
In an animal model of inflammation during pregnancy, the combination delayed the onset of labour. The experiment supplied a biological test of the model’s prediction and helped advance the combination towards human investigation. The human trial must establish its effects on pregnancy outcomes and safety; the research is not a basis for patients to start the combination themselves.[1]
SIMCO also found that folic acid, already used in pregnancy to reduce the risk of neural-tube defects, could slow immune progression towards labour. That finding illustrates another use for the system: examining how familiar medicines affect interconnected immune processes beyond the reason they were originally prescribed.[1]
Finding who is most likely to respond
The model captures differences between pregnancies as well as average responses. The researchers propose using that variation to identify people most likely to benefit from a candidate treatment. If those predictions hold in further testing, they could make trials more informative by linking a treatment to the biological pattern it is designed to change.[1]
The useful advance is a more focused route through drug repurposing: measure cell responses, connect them to clinical progression, compare candidate interventions and take the strongest lead into experiments. For this combination, the next answer will come from the people enrolled in the trial.[1]
Sources & further reading
- Digital twins of immune system in pregnancy identify drugs to prevent early labor, Stanford Medicine-led study finds Stanford Medicine · 21 Oct 2026 · Accessed 25 Oct 2026