‘Rosie, ¿Tiene Fiebre Mi Bebé?’

A University of Maryland chatbot under development to help bleary-eyed new moms manage the uncertainty of caring for a baby will soon field questions in both English and Spanish, thanks to a new grant from the National Institutes of Health.

The AI-based app called Rosie aims to bridge the divide in care for new mothers and babies of color by providing 24-hour, validated health information on topics ranging from medication dosages to major milestones and recognize signs of postpartum depression. It initially received a $3.7 million NIH grant in 2021; the additional $200,000 awarded last month will expand Rosie’s role as a reliable source for maternal and infant health advice for Spanish-speaking mothers.

“There are incredible health vulnerabilities at the intersection of color and socioeconomic status and that’s where there is the most need,” said researcher Quynh Nguyen, an associate professor of epidemiology and biostatistics. “We found that some moms just feel more comfortable asking questions in Spanish, so it was important to us to be able to offer the same experience for those moms.”

Rosie’s interdisciplinary team includes Nguyen; behavioral and community health Associate Professor Elizabeth Aparicio; computer science Associate Professor Jordan Lee Boyd-Graber; and epidemiology and biostatistics Associate Professor Xin He. They’ll begin clinical trials on Rosie in October, with Spanish capabilities available next year.

According to the Centers for Disease Control and Prevention, infant and maternal morbidity and mortality continue to be a problem in the U.S., with Black women accounting for nearly 40% of pregnancy-related deaths. Maternal mortality of Hispanic women jumped 44% from 2019 to 2020. Previous studies point to racial disparities in health care access and quality of care that can impact the frequency of well-child checkups, safe sleep practices and breastfeeding. Parents of Black children, for example, are less likely to have a routine place of care, or to believe that their pediatricians respect them as experts on their children; they are also more likely to visit the emergency room.

During a 2022 demonstration of a Rosie prototype to 109 pregnant women and new mothers of color, the researchers found that 75% of them searched for maternal and infant health information online weekly, with 90% saying they would use Rosie if available.

The researchers will continue to tap into communities to hone Rosie’s accessibility and impact, said Aparicio, particularly in helping the system recognize words and concepts that differ from region to region.

“Centering community voice and experience in the design and testing of new health technologies is critical to ensure they are addressing real mothers' needs,” she said. “We want to design interventions that not only work in a tightly controlled clinical trial, but will have broad applicability and appeal when released to the general public.”

Most AI-powered apps are in English, said Nguyen; and while most reputable sources for health information, like the CDC, offer some information in Spanish, Nguyen said it amounts to only around 30% of available English content. The boost in grant money, she said, will help researchers identify validated, Spanish-speaking health resources to add to its existing body of information from government agencies, professional medical organizations and children’s hospitals, offer culturally sensitive information relevant to different populations and help Rosie better understand and respond to Spanish-speaking moms.

While Rosie has expanded in the past year to include general health information for mothers, such as diabetes care, the next iteration of Rosie will also include anti-discrimination resources, such as where women should turn if they feel like a health care provider is prejudiced against them. Researchers hope Rosie addresses bias often associated with AI technology, which is trained with data that reflects historic inequities in socioeconomic status, race and more. But ultimately, it’s about delivering what women need, said Nguyen.

“Because of the way technology is typically designed, if it’s not conscious or intentional, it leans toward more affluent people, not these more vulnerable groups,” she said. “Sometimes women don’t know where to turn for these resources. We wanted to create something built for them first.”