In a breakthrough for digital health and wearable technology, Apple has developed an AI model that can accurately detect pregnancy with 92% accuracy, according to a new study titled “Beyond Sensor Data: Foundation Models of Behavioral Data from Wearables Improve Health Predictions.”
The model, known as the Wearable Behavior Model (WBM), leverages behavioral data gathered from Apple Watches and iPhones—including metrics like sleep patterns, heart rate variability, mobility, and more. This high-level behavioral data is sourced from Apple Health, HealthKit, and PPG-based heart rate sensors and is processed using AI algorithms trained on over 2.5 billion hours of wearable data.
Unlike previous models based solely on raw sensor data—which can be noisy and difficult to interpret—Apple’s WBM uses curated, high-level metrics validated by health experts. These include sleep quality, movement, heart rate patterns, and other physiologically relevant indicators to predict health states more accurately.
“These higher-level behavioral metrics are calculated using carefully validated algorithms… intentionally chosen to align with physiologically relevant quantities,” the study explained.
To train the AI, researchers created a pregnancy dataset from 430 pregnancies, including both vaginal and cesarean deliveries. The AI was trained to identify “positive” weeks—those during pregnancy and postpartum recovery—and compare them with “negative” weeks, when participants were not pregnant.
Additionally, data from 24,000+ non-pregnant women under the age of 50 was used to further refine accuracy and reduce bias.
The WBM not only detects the presence of pregnancy but also tracks changes in health conditions throughout pregnancy, potentially offering early insights into complications or recovery progress. As Apple Watch adoption grows, this kind of model could make passive health monitoring more powerful and personal than ever before.
This study adds to Apple’s growing focus on health-centric innovation and reinforces the potential of AI in transforming how wearable tech can monitor and predict significant health events—without the need for invasive procedures or constant clinical oversight.
As Apple continues to invest in AI-driven health features, the Wearable Behavior Model could represent a major step forward in proactive, personalized healthcare for women around the world.
