Evidence map›Paper›PMID 41502532›Full record

ArticleJAMIA open2026

Developing an artificial intelligence-powered question-and-answer chatbot with English-Spanish capabilities for new mothers.

Quynh C Nguyen, Elizabeth M Norell, Heran Mane, Xiaohe Yue, Neha Pundlik Srikanth, Carson J Peters, Francia Ximena Marin Gutierrez, Eesha Kurella, Pankaj Dipankar, Diego Salazar and 6 more

Registry-linked trialAbstract read
In one paragraph

Article in JAMIA open, 2026. The graph could read no effect estimate from its abstract, so it casts no vote on the map. It is linked to trial NCT06053515 (Rosie the Chatbot), which is not on this map. Cited by 2 papers.

0numbers the graph read from it
0cells of the map it votes in
2citing papers in PubMed
–field-weighted citation impact
1 · What the graph read from it

What it found

Each row is one number read from the abstract, on the scale the paper reported it, with its interval. Left of the dashed line favours the treatment, right favours the comparator. Under each row is the sentence it came from. New to these charts? A ten-minute tutorial.

The abstract states no effect estimate the extractor could read, or names no intervention and outcome on the map, so this paper lights no cell and moves no belief. It is still indexed, cited and linked below.

2 · The registry

The trial behind it

Trials whose registry record cites this paper, or whose number appears in the abstract. A trial that started after this paper was published is citing it as background, not reporting it.

NCT06053515 nacompletednot on this map

Rosie the Chatbot: Leveraging Automated and Personalized Health Information Communication to Reduce Disparities in Maternal and Child Health

TypeinterventionalSponsorUniversity of Maryland, College ParkRan2023 to 2026Enrolled400ConditionsPregnancy, Postpartum Depression, Infant Development, Infant ConditionsArmsRosie the Chatbot
3 · Its place in the literature

Who cites it

2 citing papers in PubMed.

  1. Article
  2. Article
4 · The record

Corrections and comments

PubMed lists nothing against this paper. Absence here is not a guarantee, only a check that was made.

5 · Who and what money

Authors and funding

16 authors.

Quynh C NguyenNational Institute of Nursing Research (NINR), National Institutes of Health (NIH), Bethesda, MD 20892, United States.
Elizabeth M NorellDepartment of Behavioral and Community Health, University of Maryland School of Public Health, College Park, MD 20742, United States.
Heran ManeDepartment of Epidemiology and Biostatistics, University of Maryland School of Public Health, College Park, MD 20742, United States.
Xiaohe YueDepartment of Epidemiology and Biostatistics, University of Maryland School of Public Health, College Park, MD 20742, United States.
Neha Pundlik SrikanthDepartment of Computer Science, UMIACS, University of Maryland, College Park, MD 20742, United States.
Carson J PetersDepartment of Behavioral and Community Health, University of Maryland School of Public Health, College Park, MD 20742, United States.
Francia Ximena Marin GutierrezDepartment of Behavioral and Community Health, University of Maryland School of Public Health, College Park, MD 20742, United States.
Eesha KurellaDepartment of Epidemiology and Biostatistics, University of Maryland School of Public Health, College Park, MD 20742, United States.
Pankaj DipankarNational Institute of Nursing Research (NINR), National Institutes of Health (NIH), Bethesda, MD 20892, United States.
Diego SalazarNational Institute of Nursing Research (NINR), National Institutes of Health (NIH), Bethesda, MD 20892, United States.
Penchala Sai Priya MullaputiDepartment of Epidemiology and Biostatistics, University of Maryland School of Public Health, College Park, MD 20742, United States.
Amrutha AlibilliDepartment of Epidemiology and Biostatistics, University of Maryland School of Public Health, College Park, MD 20742, United States.
Adwaith SanthoshDepartment of Data Science, College of Computer, Mathematical, and Natural Sciences, University of Maryland, College Park, MD 20742, United States.
Xin HeDepartment of Epidemiology and Biostatistics, University of Maryland School of Public Health, College Park, MD 20742, United States.
Jordan Boyd-GraberDepartment of Computer Science, UMIACS, University of Maryland, College Park, MD 20742, United States.
Thu T NguyenDepartment of Epidemiology and Biostatistics, University of Maryland School of Public Health, College Park, MD 20742, United States.

Funding

Sleep and Circadian Dysfunction, Brain and Neurobehavioral Development in AutismP50HD103538 · NICHD · HUGO W. MOSER RES INST KENNEDY KRIEGER · PI Stewart H Mostofsky · 2020 to 2026
$9.9M
NICHD NIH HHS P50 HD103538
6 · The paper itself

Abstract

Objectives: Generative AI chatbots are revolutionizing health education by making complex information more accessible to the public. However, their use presents risks, including bias, hallucinations, ethical concerns, and misinformation, which are particularly critical in health contexts where incorrect guidance can have serious consequences. Ensuring safety, accuracy, and reliability is essential, especially in maternal and infant health. Materials and Methods: We developed a multilingual chatbot, called Rosie, that employs a 3-stage AI pipeline, including a retriever, re-ranker, and generative model, to deliver efficient and relevant responses. To evaluate Rosie, we conducted a randomized controlled trial with pregnant and postpartum women (aged 14+ years, with infants under 6 months) from 49 US states (ClinicalTrials.gov ID NCT06053515). To analyze user interaction with Rosie, we examined 30 188 questions submitted by 197 users from October 2023 to April 2025. Six months after enrollment, a subset of Rosie participants were invited to complete a midpoint satisfaction survey assessing the chatbot's response quality, clarity, usefulness, and feature engagement. Of 105 eligible participants, 84 completed the survey (80% completion rate) between November 2024 and June 2025. Results: Users asked an average of 2.73 questions per day, with increased activity on weekdays and evenings, peaking at 10 p.m. The most highly rated topics included infant feeding, developmental milestones, symptoms (eg, fever), and hospital/birth preparations. Midpoint feedback was positive: 86% rated the answers as high quality, 91% found them useful, 95% found them easy to understand, and 84% expressed satisfaction. Discussion: In the initial phase, Rosie users reported technical issues and less satisfactory responses. However, integrating a retrieval-augmented generation system, expanding Rosie's knowledge base, and adding more interactive features led to sustained increases in positive ratings and consistently high user satisfaction. The Spanish-language expansion, enabled by a multilingual pipeline and advanced translation models, directly addressed pilot study feedback and further broadened Rosie's accessibility. Rosie offers a broad approach, supporting users through pregnancy, childbirth, postpartum, and infant care during the first year. Conclusion: Chatbots like Rosie have the potential to transform health information delivery by providing a scalable, personalized, and cost-efficient solution.

Indexed as

artificial intelligencechatbotchild healthhealth informationmaternal health

Identifiers

PMID41502532
PMCPMC12772640

What OpenQuestion holds

Textmetadata
LicenceCC BY-NC
Read underepoch 390

Registered trials

Read under generation 80e0d062 · epoch 390. Bibliography from PubMed, PubMed Central and OpenAlex; grants from NIH RePORTER; trial links from ClinicalTrials.gov; estimates, votes and beliefs from the OpenQuestion graph.