Stanford's Virtual Pregnancy Model Predicts Which Drugs Could Prevent Preterm Birth Without Human Trials
A digital twin trained on immune cell data identified two drug candidates that could extend pregnancy by up to 24 days, offering a solution to the ethical dilemma of testing therapies in pregnant women.
Preterm birth affects one in ten pregnancies, yet developing treatments has remained nearly impossible due to the ethical constraints of conducting clinical trials in pregnant populations. Now, Stanford University researchers have created a digital twin model that simulates immune responses during pregnancy, potentially unlocking new prevention strategies without exposing expectant mothers to experimental interventions.
The technology, called SIMCO (simulated immune modeling of clinical outcomes), analyzes single-cell immune protein data to predict whether a pregnancy will end in preterm or term labor—and crucially, to test how different drugs might alter that outcome. Published in Science Advances, the research identified two promising drug candidates that could extend gestation by modulating immune pathways associated with labor initiation.
Building a Digital Pregnancy at Single-Cell Resolution
The Stanford team, led by proteomics researcher Brice Gaudillière, collected blood samples from nonpregnant individuals and second-trimester pregnant women, exposing them to inflammatory stimulants. Using mass cytometry, they analyzed 21 proteins and activation markers across 28 immune cell populations, identifying 4,480 immune activation features.
SIMCO was trained on 13 markers from a previous clinical study of gestational outcomes. The model successfully discriminated between term and preterm samples and predicted actual delivery timelines, creating what researchers describe as a virtual testing ground for therapeutic interventions.
The researchers then tested nine candidate drugs through the digital twin platform. Most suppressed immune activity, with naive CD4+ and CD8+ T cells and central memory CD4+ T cells most commonly affected by the interventions.
Two Drug Candidates Emerge from Virtual Testing
Three interventions showed particular promise in extending time to labor: maprotiline, tetrahydrofolate, and a combination of salicylic acid with lansoprazole. Each predicted an extension of 17.3 to 24.7 days, according to the model. From these, SIMCO identified two potential drugs that could prolong gestation to prevent preterm labor.
Tetrahydrofolate and the salicylic acid/lansoprazole combination appeared to work through shared mechanisms—suppressing activating pathways in naive CD4+ T cells while simultaneously increasing immunosuppressive protein expression in granulocytes. This dual action on the immune system could explain their potential to delay labor onset. "So, it all makes sense biologically. That was super exciting," Gaudillière said in a press release.
Most cases of preterm birth occur spontaneously without known risk factors, making prevention particularly challenging. The ability to identify existing drugs that could be repurposed addresses both the scientific and ethical barriers that have stymied progress in this field.
Implications for Maternal-Fetal Medicine
For healthcare providers, SIMCO offers a potential pathway to evidence-based preterm labor prevention strategies without the traditional requirement for large-scale clinical trials in pregnant populations. If validated, the approach could reduce neonatal intensive care costs and improve infant outcomes by preventing premature deliveries.
The technology also opens doors for health AI companies developing similar single-cell resolution digital twin platforms for other conditions where clinical trials pose ethical or practical challenges. Pharmaceutical companies could use this approach to accelerate drug repurposing efforts across vulnerable populations.
Stanford University has filed a patent related to this work, signaling potential commercialization of the platform. As digital twin technologies mature, they may fundamentally reshape how medicine develops treatments for populations that cannot safely participate in traditional clinical research.
Sources
Google Search
Follow Health AI Central on Google
Choose Health AI Central as a preferred source, and Google is more likely to show our reporting in Top Stories.
Reporting & accountability
How we maintain accuracy
Health AI Central separates reported facts, analysis, and public-source data. Readers can review our standards, request a correction, or consult the relevant research methodology before relying on a claim.