New AI Model Targets Improved Breastfeeding for NICU Babies
A new artificial intelligence model developed at the University of Florida College of Medicine aims to enhance breastfeeding rates among mothers of premature infants in the neonatal intensive care unit (NICU). This initiative, known as the Maximizing Initiatives for Lactation Knowledge (MILK+) project, is designed to support families during a critical time when breastfeeding can be particularly challenging.
The AI-driven approach combines the expertise of clinicians, nurses, and AI engineers at UF Health Shands Hospital in Gainesville. The goal is to identify mothers who may struggle with lactation and to implement strategies before they give birth. This proactive measure is especially crucial for mothers of preterm infants, who often face unique challenges in establishing breastfeeding.
Dr. Helen Hu, the project lead and a clinical assistant professor in the Department of Pediatrics, emphasized the significant benefits of breastfeeding. “Breastfeeding has so many potential benefits for both mother and child,” she noted. Research indicates that breastfeeding aids maternal recovery and lowers the risk of conditions such as high blood pressure and certain cancers. For infants, especially those born prematurely, maternal breastmilk supports healthy weight gain and brain development while reducing the likelihood of illnesses.
Despite a high initiation rate of breastfeeding in Florida—approximately 85%—the duration of breastfeeding tends to decline significantly over time, according to data from the Florida Department of Health. The MILK+ project seeks to address this gap by integrating AI models that assist healthcare teams in predicting breastfeeding difficulties.
AI Models Enhance Lactation Support
The AI models focus on two key intervention points: prenatal and postnatal. The prenatal model synthesizes patient data to identify risk factors, including preexisting conditions and socioeconomic indicators. In contrast, the postnatal model utilizes data on breastmilk production during the hospital stay to forecast whether a mother will successfully continue breastfeeding upon discharge.
Data from over 18,000 mothers and more than 22,000 newborns over several years informed the development of these models. Dr. Tanja Magoc, an associate director of the AI/Quality Improvement team, explained the collaborative effort involved in refining these predictive tools. “The AI model helped us screen which pieces of information were important indicators of the development of maternal breastmilk,” she said.
Both models have demonstrated impressive accuracy, with the postnatal model achieving a 95% accuracy rate in predicting lactation success. The insights generated will enable clinicians to identify at-risk patients and provide timely interventions, such as early consultations with lactation specialists.
The lactation support committee at UF Health Shands Children’s Hospital includes trained nurses, lactation consultants, and neonatologists dedicated to assisting breastfeeding mothers. The hospital also holds the Baby-Friendly designation from Baby-Friendly U.S., a global initiative by the World Health Organization and UNICEF, acknowledging facilities that promote effective breastfeeding practices.
Future Directions and Considerations
As the AI models continue to evolve, the team aims to refine their predictions and expand their use across other UF Health facilities. Dr. Hu emphasized the importance of collaboration between clinical staff and AI developers to ensure the models provide actionable insights.
While the models offer valuable predictions, Dr. Hu cautioned against over-reliance on AI. “The model is only as good as the data it was trained on,” she stated. Acknowledging that certain patient populations may be underrepresented, she urged clinicians to apply their judgment in conjunction with AI insights.
The MILK+ project represents a significant step forward in addressing the complexities of breastfeeding for mothers of NICU infants. By harnessing the power of AI, healthcare teams can enhance lactation support, ultimately benefiting both mothers and their vulnerable newborns. As this innovative approach continues to develop, it holds the potential to transform care practices in maternity and neonatal health across the region and beyond.
