The technology for tracking a baby’s heartbeat and vital signs during birth works much the way it did when it was developed in the late 1960s. As a result, two clinicians reviewing the same tracing from a fetal electronic monitor will agree only 20% to 40% of the time. That uncertainty is one reason the United States has among the highest rates of maternal and infant morbidity and mortality of any wealthy nation.
Dartmouth researcher Tam Vu suspects that artificial intelligence can do better. He envisions wearable devices and integrated software that take the guesswork out of childbirth by monitoring the mother and child synchronously and telling the care team what it sees—and what to do next.
Professor Tam Vu
Vu, the Thomas A. and Georgina Tugwell Russo 1977 Distinguished Professor in Computer Science, will work to build that system as the lead principal investigator and technical lead for HARMONI, a new project recently funded with a one-year Phase I contract award of up to $5 million by the Advanced Research Projects Agency for Health, or ARPA-H, an agency within the U.S. Department of Health and Human Services.
HARMONI—short for Hypoxia Assessment via Real-time Maternal-fetal Oxygenation and Neurophysiological Integration—treats a mother and child as a coupled system linked by a wireless maternal abdominal smart belt and a companion maternal earpiece that monitors both of their vital signs. An AI inference layer will give clinicians objective, real-time information and clear, cause-specific guidance during labor.
Vu will lead the project with his longtime collaborator at Columbia University, Xia Zhou, a computer scientist and the co-lead principal investigator who will direct the development of the abdominal belt, the AI and machine learning work, and clinical coordination at Columbia. The HARMONI team is one of eight research teams to receive contract awards in the first year of ARPA-H’s four-year Making Obstetrics Care Smart program, known as MOCS, which aims to make the U.S. the safest place in the world to have a baby.
“It’s a true honor to be entrusted to build this platform, from the hardware to the integration, into a backbone for maternal care,” Vu says. “For the first time, we’ll bring mother and child together through continuous, real-time tracking. Our vision is to improve health outcomes by deploying intelligent sensing technologies with long-lasting, life-changing results.”
‘Evidence rather than guesswork’
About 4 million babies are born in the United States each year. Despite the United States spending more per capita on maternal care than any other country, about 83,000 unnecessary cesarean sections are performed each year due to a lack of reliable data on fetal health. The outcome is more than $1 billion in unnecessary health costs and a mortality rate affecting Black and Indigenous mothers at two to three times the rate of white mothers.
With HARMONI, a wireless belt worn on the mother’s abdomen monitors uterine contractions along with fetal heartbeat and motion via an integrated ultrasound module. At the same time, a small sensor worn on the mother’s ear tracks her own vital signs. The AI combines the two data streams and weighs them against each other to give the care team an objective and comprehensive read of maternal-fetal health in the moment, as well as a prognosis of the underlying cause and suggestions for next steps.
Whereas now two clinicians might study the same output from a fetal electronic monitor and come to different conclusions, HARMONI is built to deliver one consistent answer, Vu says. Better information could mean fewer cesareans that turn out to be unnecessary, and faster action when a baby is genuinely in distress.
"For decades, labor and delivery decisions have rested on a 1970s-era heart rate strip and clinician intuition," Vu says. "HARMONI brings real-time, multi-sensor data into the room. Our goal is to turn one of the most consequential moments in medicine into a decision made with evidence rather than guesswork."
The ear sensor is familiar territory for Vu. He coined the term “earable” for computing that reads the body’s signals from the ear.
The award’s first phase focuses on the project’s technical backbone: the belt and ear sensor, the software that fuses their signals, and an early conversation with the U.S. Food and Drug Administration about a path to approval, along with a small feasibility study. Later phases could carry the system toward an FDA submission.
The work will involve several institutions. Phuc Nguyen at the University of Massachusetts Amherst leads the engineering that links the wearable sensors to hospital systems. Matthew Hoffman and William Fifer at Columbia provide clinical leadership, and John Yeh, professor emeritus at Harvard Medical School, advises on clinical strategy. The team plans to test the system across three settings: Columbia University Irving Medical Center, Geisinger, and the Gentle Landing Birth Center.
This research was funded, in part, by the Advanced Research Projects Agency for Health (ARPA-H). The views and conclusions contained in this document are those of the authors and should not be interpreted as representing the official policies, either expressed or implied, of the U.S. Government.