Apple Watch detects pregnancy with 92% accuracy in experiment

Years ago, on a flight from the United States to Brazil, 50-year-old lawyer Blima Katz fastened her seatbelt and blacked out for 12 hours straight. "Even so, when I got home, I slept through the night. The next morning, I remembered how sleepy I'd been during my first pregnancy. I thought I'd take a pregnancy test," she recalls. Another symptom that made Blima suspicious was her heightened emotions, which made her cry over anything, even during vacation. "The test confirmed the pregnancy, indeed, before her missed period," says Blima, who is now the mother of two teenage girls.

It's possible that in the near future, women will have another ally in determining whether they're pregnant: their smartwatches. Recently, in partnership with the University of Southern California, Apple developed an artificial intelligence model called the Wearable Behavior Model (WBM) that uses behavioral metrics from the Apple Watch to predict health conditions. In total, approximately 2.5 billion hours of data were generated by approximately 162,000 participants (men and women) over a five-year period (2019 to 2025).

By assessing changes in walking pace, sleep duration, heart rate fluctuations, and step count, VBM was able to identify early signs of a variety of conditions, from infections to cardiovascular problems and even pregnancy. In the case of pregnancy, the accuracy was 92%. Could this type of technology soon influence medical protocols or diagnoses? For gynecologist and obstetrician Carol Dalboni, technical director of Clínica Dedicali (RJ), the odds are high. And they require caution.

" One of the benefits, in my opinion, is that the earlier you start medical protocols for pregnancy, such as taking vitamins, the better. This is also true for removing teratogenic substances (that are, substances that affect fetal development). It's very common, for example, for women to ingest alcohol or take certain medications because they didn't know they were pregnant," explains the specialist. In the case of women with irregular cycles, it can take months for them to realize it and, during this time, may continue habits that are harmful to the health of both mother and baby.

On the other hand, Dalboni fears the legal implications of this type of application regarding the data protection of pregnant women. One risk would be sharing data with third parties, from health plans to employers. "What if the application is paired with a computer at the company where the pregnant woman works, and the company has access to her data?" he reflects.

Another important discussion raised by the Apple experiment was the comparison with traditional pregnancy tests. The most common test today, according to gynecologist and obstetrician Carol Dalboni, is the beta hCG test, which measures the level of the hormone human chorionic gonadotropin (hCG) in the blood. This hormone is a marker of pregnancy, and the amount present in the body can not only confirm pregnancy but also indicate certain pathologies. "If the level is too high, for example, incompatible with the gestational age, it could indicate a molar pregnancy (a condition in which abnormal growth of placental tissue occurs)," explains the specialist.

Furthermore, the accuracy of Apple's experiment was 92%, as mentioned above, while the beta hCG test was 99%. "For these reasons, I believe this type of technology can be an adjunct to, but not a replacement for, conventional testing," he concludes.