Eight years ago, Kenneth J. Moise, MD, a maternal-fetal medicine specialist at Dell Medical School, The University of Texas at Austin, was on a phone call with his wife when she heard a loud rumble from his end of the line.

“I had her on speaker with the phone on my abdomen” Moise said. “She thought it was storming but I looked down and realized she was hearing the coffee in my belly. I thought to myself: ‘If the phone’s microphone is sensitive enough to pick up intestinal movement, couldn’t it pick up a baby moving?’”

Moise tested the theory on his daughter when she was pregnant with her first child. Each week for about 6 weeks, she would place her phone on her stomach, record an audio strip, and then write down when she felt fetal movements on a piece of paper.

Moise sent the recordings to researchers at the Texas Advanced Computing Center at The University of Texas at Austin, which marked the beginning of his journey to determine if smartphone-recorded audio could reliably detect fetal movement. 

Decreased fetal movement has been associated with stillbirth, which is classified as the death of a fetus after 28 weeks of pregnancy. As of the end of 2025, the US stillbirth rate is approximately 1 in 175 pregnancies or around 20,000 pregnancies annually.

Moise led a study presented at the Society for Maternal-Fetal Medicine (SMFM) 2026 Annual Pregnancy Meeting and published in Obstetrics & Gynecology that analyzed 136 pregnant women who were recruited by research nurses in community-based obstetric clinics and inner city prenatal clinics in Austin. The median age was 32.6 years.

They included women who spoke English or Spanish, were 18-45 years of age, and carried a single baby with pregnancy dated by a first-trimester ultrasound. Patients were between 24 and 38 weeks and 6 days of pregnancy, had no medical complications, and their baby showed normal growth with no congenital anomalies.

Researchers assessed three kinds of movement: gross fetal movement, defined as a kick or roll; breathing movements; and hiccups. Fetal movements were recorded using a continuous ultrasound, which watched the baby over the course of 30 minutes, smartphone audio recordings, and maternal reporting. Each patient’s perception of movement was recorded through a laptop where they pushed a button each time they felt movement. 

Two sonographers reviewed video offline to log precise timing of movements. Patients were not allowed to watch the ultrasound screen to avoid bias. 

Audio recordings detected gross fetal movement with 88.6% accuracy (95% CI, 88.3%-88.8%) compared to maternal perception of fetal movements (18%), fetal breathing movements with 91% accuracy compared to maternal perception (3%; AUC, 91%; 95% CI, 90.6%-91.1%), and fetal hiccups with 68% accuracy (95% CI, 65.6%-71%; P < .001), compared to maternal perception (32%). 

“We’re two, three times better at picking up movements just by holding the iPhone to your abdomen and letting the iPhone pick up the movements,” Moise said.

Rene Cortese, PhD, associate professor in the department of internal medicine, University of Kansas Medical Center, said the new approach could improve remote monitoring of fetal well-being. 

“Although it will not replace the regular obstetric checkups, I anticipate that the detection system can provide real-time access to the baby’s movement,” Cortese said. “By accessing realtime data, the health provider may monitor abnormalities in fetal movement without the need for in-person meeting. On the other hand, it may reduce anxiety for the mother throughout pregnancy during the inter-checkup intervals.” 

Moise said the findings highlight a larger issue in prenatal care: There is a lack of education around fetal movement among both physicians and patients.

“We’re not educating our OBs how to talk to patients about fetal movement,” Moise said. “Most only ask if the baby is moving — not ‘has your baby’s strength of movement changed’ or ‘how is your baby moving?’”

He added that AI could help modernize monitoring. His plans are to expand the research into a smartphone app that allows pregnant patients to record fetal movements at home and update their doctors in real-time.

“I think the day is going to come where we don’t have to ask patients to come in twice a week to be put on the monitor,” Moise said. “Instead, we’re going to say, ‘Send me your data from your wearable,’ and the AI system is going to say, ‘Hey, patient X has got a movement alarm signal going on. You need to contact her and get her into the hospital to evaluate the baby, see if it needs to be delivered.’ That’s where it’s going. AI will help us with that.”

Moise and Kelly Gaither are co-inventors on patent applications submitted through The University of Texas at Austin Discovery to Impact office: detecting and classifying fetal movements using machine learning estimation of audio recordings. No other financial disclosures were reported.