A new handheld device with artificial intelligence (AI) can predict how far along a pregnancy is as accurately as experienced ultrasound technicians, according to a new study.
The study tested this AI-powered device on 400 pregnant women in Zambia and North Carolina. It found that the device’s predictions were only slightly off by an average of 3.2 days, compared to the experts’ 3.0 days. This small difference shows that the AI device is just as reliable.
Researchers found that the AI device correctly estimated the due date within a week for about 91% of the cases, almost as good as the 93% accuracy of traditional methods. Dr. Jeffrey Stringer from the University of North Carolina led the study, which was published in JAMA.
Dr. Stringer mentioned that this AI tool is especially useful in places where it’s hard to get traditional ultrasound equipment. Handheld devices are cheaper and need less training to use. This means more pregnant women can get accurate care without traveling far.
In a previous study, Dr. Stringer showed that this AI tool could predict due dates just as well as trained experts, even in areas with fewer resources.
This new study confirms that the AI tool works well. It’s different from AI like ChatGPT, focusing on specific tasks. The research team is also working on adding more features to the device.
Dr. Stringer believes that in the future, patients won’t have to travel long distances for an ultrasound. Instead, they can get the same quality of care at local clinics.
In an editorial, Dr. Methodius Tuuli and colleagues noted that such tools could help improve pregnancy care in low-resource settings. They pointed out that while this AI tool is a big step forward, it’s not a complete solution yet.
The study involved 400 women with single pregnancies. Women with very high body weight, twins, or known fetal issues were not included.
The average age of the mothers was 29 years, and the average time of the first ultrasound was at 11.7 weeks of pregnancy. Some women were HIV positive, and many had given birth before.
After experts established the correct due date using traditional methods, novice users with no prior training used the AI device to estimate the due date. These new users received just one day of training.
The study found that the AI device’s predictions were almost as accurate as those of trained experts. However, the study didn’t include high-risk pregnancies, and more testing in different locations is needed.
Dr. Tuuli and colleagues also mentioned that the study had a small sample size and limited training details for the new users.
Source: medpagetoday
