Wearable tech data shows promise in ADHD detection
Using Fitbit data from the the largest long-term study of brain development and child health in the United States, researchers employed machine learning to test whether physiological markers could accurately predict ADHD diagnoses.
A new study published in Frontiers in Child and Adolescent Psychiatry reveals that common wearable devices like Fitbits may hold the key to improving how we identify Attention-Deficit/Hyperactivity Disorder (ADHD) in adolescents. By analyzing patterns in heart rate, activity levels and energy expenditure, researchers were able to predict ADHD diagnoses with striking accuracy, offering a glimpse into a future where objective, real-time data supports earlier and more personalized mental healthcare.
A fresh approach to a common challenge
ADHD affects approximately 1 in 10 children and adolescents in the United States. It is typically diagnosed based on parent and teacher reports, clinical interviews and behavioral observations. While effective, these methods rely heavily on subjective interpretation and can sometimes miss important nuances in how symptoms appear over time. This study, led by Muhammad Mahbubur Rahman, PhD, and colleagues at Children’s National, sought to determine whether wearable health data could help fill that gap.
Turning Fitbit metrics into meaningful insights
The study used data from 450 adolescents who were part of the larger Adolescent Brain Cognitive Development (ABCD) study, the largest long-term study of brain development and child health in the United States. Each participant wore a Fitbit, which captured three key activity and physiological measures:
- Resting Heart Rate (RHR) – the number of heart beats per minute while the body is at rest
- Sedentary Time – time spent with little or no physical activity
- Energy Expenditure – estimated calories burned through physical activity
When the researchers compared these measures between teens with and without ADHD, they found statistically significant differences. Teens with ADHD had consistently higher resting heart rates and showed distinctive patterns in both their movement and stillness.
To go further, the team applied a machine learning model to test whether these physiological markers could accurately predict ADHD diagnoses. The model performed extremely well with 89% accuracy, 88% precision, 90% recall and a 0.95 area under the curve (AUC). These results suggest that the combination of passive, continuous data and predictive modeling could serve as a valuable screening tool, particularly in settings where full clinical evaluations are difficult to access.
A path toward more accessible mental healthcare
The implications are big. If validated in larger and more diverse populations, wearable-derived data could offer a low-cost, low-burden way to flag teens who might benefit from further ADHD evaluation. This could lead to earlier support, fewer misdiagnoses and more tailored treatment strategies.
Importantly, this approach isn’t about replacing clinicians, it’s about giving them better tools. Real-world, real-time data from wearables could act as an additional layer of insight that supports more precise, individualized care. As wearable technology becomes more embedded in daily life, its role in healthcare, especially adolescent mental health, is poised to grow.
You can read the full study, Unlocking the potential of wearable technology: Fitbit-derived measures for predicting ADHD in adolescents, in Frontiers in Child and Adolescent Psychiatry.
