1 in 10 births in the U.S. are premature. Here’s how AI could help doctors predict it

Key Concepts

• Premature Birth: Birth occurring before 37 weeks of pregnancy.
• NICU (Neonatal Intensive Care Unit): A specialized unit for newborns requiring intensive medical care.
• AI (Artificial Intelligence): Technology that enables computers to perform tasks that typically require human intelligence, such as learning and problem-solving.
• Ultrasound Images: Medical imaging technique using sound waves to create images of internal body structures.
• Ellis van Creveld Syndrome: A rare genetic disorder affecting bone growth, teeth, and heart development.
• FDA (Food and Drug Administration): U.S. agency responsible for protecting public health by ensuring the safety, efficacy, and security of human and veterinary drugs, biological products, medical devices, etc.

Experiences of Parents of Preterm Babies

The video begins by highlighting the significant issue of premature births in the U.S., with one in 10 babies born before 37 weeks of gestation, a rate considered high among developed nations. The March of Dimes facilitated connections with parents who shared their deeply personal and challenging experiences with preterm births.

• Erica Shoemate (Nia's Mom): Her daughter, Nia, was born with Ellis van Creveld syndrome, a very rare genetic condition with fewer than 300 reported cases globally. The rarity was compounded by their family being of color, with only two other known families of color impacted worldwide. Nia is now a spunky toddler approaching her third birthday.
• Roxanne Romeo (Zara's Mom): Zara was born at 26 weeks and 4 days, weighing 2 pounds, 6.7 ounces. She spent 93 days in the NICU. Roxanne describes the daily anxiety of calling to check on Zara's breathing episodes, where the baby stops breathing. She emphasizes that despite the darkness, perseverance leads to finding a light, even if it's not as imagined.
• Lucero Payano (Zayel's Mother) and Ruban Contreras (Zayel's Dad): Zayel was born at 25 weeks and 2 days. He is now six years old, in first grade, and uses a communication device. He is described as a "little rock star" who is doing "amazing," defying expectations for a 25-weeker. He enjoys playing tricks on his teachers.
• Ashley O'Neil (Colin's Mom): Ashley went into labor at 25 weeks with Colin. She was at a small community hospital lacking the necessary resources, and Colin was born weighing approximately 1.5 pounds. He spent 183 days in the NICU, fighting and repeatedly pulling out his tubes, but ultimately came home.
• Emma Lee Greenley (Lois' Mom): Lois was born at 24 weeks and 2 days, weighing 1 pound, 7 ounces. She was immediately intubated. Lois is now five years old, in preschool, and doing very well.

These parents offer encouragement to others navigating similar situations, advising them to allow themselves to grieve the pregnancy and parenthood experience they didn't have, and to avoid comparing their child's journey to others. They stress the importance of both parents being informed about prematurity.

AI for Predicting Premature Births

The video then introduces Robert Bunn, an entrepreneur whose personal experience with his wife's nine miscarriages inspired him to develop technology to predict premature births.

The Problem: Surprise Premature Births

Robert Bunn explains that a primary reason for the high rates of premature births is that they are often a surprise. Doctors typically lack forewarning, making it difficult to intervene effectively. Traditional screening methods for prematurity are challenging because:

• Ultrasound images are "noisy" and hard to read: It's difficult to discern subtle issues.
• Known risk factors apply to a small percentage of women: Many women with a history of premature birth or miscarriage are monitored closely, but the accuracy of prediction is limited.

The Solution: AI-Powered Ultrasound Analysis

Bunn's solution involves using AI to analyze millions of ultrasound images and correlate them with the actual outcomes of those pregnancies.

Methodology:

1. Data Acquisition: Obtain millions of ultrasound images.
2. Outcome Mapping: Record the outcomes for the babies in those images (e.g., whether they were born prematurely).
3. AI Training: Create an AI model that learns to map patterns within the pixel data of the ultrasound images to the recorded outcomes. This required building a supercomputer from parts in his basement due to the immense computational power needed.
4. Pattern Recognition: The AI was trained over months to identify subtle patterns in the images that correlate with specific pregnancy outcomes.
5. Prediction Capability: This process enabled the AI to predict delivery dates with greater precision and identify potential issues leading to premature birth.

Bunn emphasizes that the goal is to provide doctors with precise delivery date predictions, allowing them to act confidently and treat at-risk pregnancies more effectively.

Real-World Applications and Results

Bunn's technology, "Ultrasound AI," is currently in trial use in Brazil and Chile. The results have been highly positive:

• Identification of First-Time Mothers at Risk: The AI can identify women having their first child who may be at risk of premature birth, ensuring they receive necessary care they might otherwise miss.
• Improved Management of High-Risk Pregnancies: For high-risk pregnancies where keeping the baby in utero longer is beneficial, the AI provides a good estimate of spontaneous delivery timing. This allows doctors to potentially schedule a C-section beforehand, optimizing delivery timing for better outcomes and reducing NICU stays.

Outlook for the U.S.

The technology has recently undergone final review for FDA approval and is expected to be available in the U.S. within the next few months. Bunn expresses confidence in the technology's ability to save babies' lives in America.

Synthesis and Conclusion

The video powerfully illustrates the profound impact of premature births on families, showcasing the resilience and hope of parents who have navigated this challenging journey. It then pivots to a technological solution, highlighting how AI, through sophisticated analysis of ultrasound images, offers a promising new approach to predicting premature births. By identifying subtle patterns invisible to the human eye, this AI can equip medical professionals with crucial information to intervene earlier, improve outcomes, and ultimately save lives. The successful implementation in South America and the impending FDA approval signal a significant advancement in maternal and neonatal care.