Generative Artificial Intelligence and Maternal Health Outcomes in Pregnant and Postpartum Individuals: A Randomized Clinical Trial.
Elizabeth M Norell, Xiaohe Yue, Heran Mane et al.
✦ AI-curated · Sources linked
In 30 seconds
This 2-group randomized clinical trial assessed whether a generative AI maternal health chatbot improved outcomes among 400 pregnant or postpartum participants from racial and ethnic minoritized groups recruited across the United States. Participants received chatbot access or monthly children’s books as an active control, with outcomes assessed at 12 months. The chatbot was not associated with improved anxiety or depression scores: adjusted mean differences were 0.10 (95% CI, -0.77 to 0.96) and -0.23 (95% CI, -1.05 to 0.59), respectively. Most other measured outcomes also did not differ between groups.
Key findings
- At 12 months, anxiety scores did not differ between groups (adjusted mean difference, 0.10; 95% CI, -0.77 to 0.96).
- Depression scores did not differ between groups (adjusted mean difference, -0.23; 95% CI, -1.05 to 0.59).
- No differences were reported in emergency department visits, postpartum care, contraception use, or sleep duration.
- More self-reported pregnancy-related complications appeared in a secondary endpoint analysis, but not in longitudinal analyses.
Why it matters
Randomized evidence can help distinguish the promise of digital maternal health tools from demonstrated benefit. In this trial, chatbot access did not improve anxiety or depression at 12 months compared with an active control, and most secondary outcomes were also similar. The findings underscore the importance of evaluating both intended benefits and potential safety signals when assessing AI-supported maternal health resources.
What to keep in mind
The pregnancy-complication finding was present in a secondary endpoint analysis but not longitudinal analyses, and should be interpreted cautiously. Higher-engagement findings were exploratory within-group associations.