A randomized controlled trial published in JAMA Network Open on November 4, 2026 found that Rosie, an AI‑driven chatbot designed to support maternal and infant health, did not produce a measurable improvement in anxiety or depression after 12 months compared with a control group that received a monthly children’s board book.

Study design and population

The trial recruited 400 English‑speaking participants aged 14 or older who were either pregnant or parenting an infant younger than six months. All participants identified as members of racial or ethnic minoritized groups in the United States. Enrollment ran from October 2023 through December 2024, with follow‑up completed in December 2025.

Intervention and control conditions

Participants were randomly assigned to either Rosie access (200 users) or a control condition (200 users). Rosie provided answers to health questions, personalized guidance, reminders and educational content, while a clinical social worker monitored distress alerts and offered resources. The control group received one children’s board book each month for a year. Both arms could communicate with study staff and received modest incentives for engagement.

Primary outcomes and results

Self‑reported anxiety (GAD‑7) and depression (PHQ‑9) scores were the primary endpoints. At 12 months, 361 participants contributed anxiety data and 362 contributed depression data. The adjusted mean difference for anxiety was 0.10 points (95 % CI ‑0.77 to 0.96) and for depression was ‑0.23 points (95 % CI ‑1.05 to 0.59). Both confidence intervals crossed zero, indicating no statistically significant benefit from Rosie.

Interpretation and limitations

Researchers noted that baseline symptom levels were low (median anxiety score 2, median depression score 3) and that self‑reporting and the added human support could limit the applicability of findings to routine chatbot use. The study therefore cannot confirm equivalence between Rosie and the book program, nor can it rule out benefits in different populations or with alternative AI designs.

What to watch next

Future research may explore AI‑based maternal health tools in higher‑risk cohorts, assess longer‑term outcomes, or evaluate chatbot performance without supplementary human oversight. Stakeholders will be watching for subsequent trials that could clarify the role of digital assistants in perinatal mental‑health care.

FAQ

What was the primary comparison in the Rosie trial?

The trial compared access to the Rosie AI chatbot with a control condition that provided a monthly children's board book for one year.