Stanford Medicine researchers have used digital models of the immune system to identify a drug combination for testing against preterm labour. Pairing aspirin with lansoprazole produced a much larger predicted delay than aspirin alone in the team's simulations. Supporting animal work has led the researchers toward a human trial, Stanford reported on 21 October.[1]
The system, SIMCO—simulated immune modeling of clinical outcomes—connects laboratory measurements with the changing immune patterns of pregnancy. Its role is to rank treatment candidates before a clinical trial. The most eye-catching result, a fifteenfold increase in the simulated delay relative to aspirin alone, belongs to that computational comparison; it does not describe delivery dates or benefits observed in pregnant patients.[1]
Building a model of the immune clock
Pregnancy involves a changing relationship between immune tolerance and inflammation. The researchers studied patterns that move with this progression, seeking signals that help explain why labour begins when it does. A model able to represent those changes could test whether a proposed intervention moves the immune system toward or away from the pattern associated with delivery.[1]
The team measured responses from roughly a quarter of a billion immune cells exposed to medications. It combined those measurements with repeated observations from 63 pregnancies and their known outcomes. This gave the virtual models both information about drug effects on cells and a longitudinal account of immune changes during pregnancy.[1]
Stanford reports that SIMCO could predict the timing of labour in the study's evaluation. That finding supports the usefulness of its immune signatures. Selecting a treatment adds a further causal question: whether deliberately changing those signatures improves pregnancy outcomes. The trial is intended to move the research from a predictive association toward that clinical test.[1]
Why the pairing stood out
Researchers compared nine regimens drawn from medicines with existing pregnancy-related safety information. Aspirin and lansoprazole emerged as a promising combination, with the model suggesting complementary effects on inflammatory processes. Lansoprazole is used to treat acid reflux; the proposed use here concerns a different biological effect and a different treatment goal.[1]
From model to clinical study
| Stage | Evidence reported |
|---|---|
| Cell experiments | About 250 million immune-cell responses supplied information about medication effects |
| Pregnancy observations | Longitudinal data from 63 pregnancies connected immune patterns with outcomes |
| Virtual screening | Nine regimens compared; the combination produced a much larger predicted delay than aspirin alone |
| Animal experiment | The combination delayed labour in an inflammatory pregnancy model |
| Human trial | Stanford reports FDA permission and work to begin clinical testing |
What human testing needs to establish
The reported animal result gives the treatment hypothesis a second kind of support beyond the simulation. Human testing must now determine whether the pairing is safe and beneficial for the patients enrolled, including outcomes for both mother and baby. Existing experience with the individual medicines does not establish the effectiveness of their combined use to prevent preterm birth.[1]
The model also draws on a relatively small number of tracked pregnancies. Its usefulness will depend on how well the immune patterns carry across patients with different histories and causes of early labour. A detailed immune model captures one part of a pregnancy; its predictions will need to be evaluated alongside the broader clinical picture.[1]
SIMCO's contribution could extend beyond this particular combination. By connecting individual immune profiles with experimental drug responses, it offers a way to narrow a large set of possibilities into a smaller set of testable hypotheses. The next decisive evidence will be the human trial's measured outcomes, which will show whether the treatment selected by the model produces the benefit it predicts.[1]
Sources & further reading
- AI twins find drugs that may prevent early labor Stanford University / Stanford Medicine · 21 Oct 2026 · Accessed 27 Oct 2026