Publications

This study explores a faster, lower-cost method for identifying internalizing disorders (like anxiety and depression) in young children. Instead of relying on hours of interviews, the researchers used a 90-second task designed to induce mild fear, while monitoring children’s movement with a wearable sensor. Using machine learning on just 20 seconds of movement data, the model predicted diagnoses with 75% accuracy—comparable to traditional assessments, but significantly more efficient. This research demonstrates the potential for wearable-based screening tools in clinical settings.

This study tested whether a simple, wearable motion sensor could detect different phases of threat response—such as anticipation, startle, and recovery—in young children. Findings from 18 participants (ages 3–7) showed that specific movement patterns varied by threat phase and were linked to anxiety-related family and symptom factors. The results suggest motion sensing may offer a feasible, objective method for studying early emotional reactivity in mental health research.

This study introduces the ChAMP System—a mobile app and open-source platform for collecting movement and audio data during mood-based tasks to explore childhood mental health. Data from 101 children showed that machine learning models could detect emotional and behavioral concerns with accuracy comparable to parent-report tools. The system aims to make digital biomarker research more accessible and clinically relevant.

This study builds on prior work to further validate a rapid, wearable-based method for identifying anxiety and depression in young children. Using motion data collected during a brief 90-second task, the researchers tested a clinically practical 20-second window and applied logistic regression models to predict diagnoses. The approach achieved 80% accuracy—matching traditional diagnostic tools while requiring far less time and cost. The findings reinforce the potential of wearable sensors and machine learning to support faster, more accessible screening in pediatric care.

This study highlights how objective, sensor-based measures—such as movement and physiological signals—can identify broad patterns of childhood mental health impairment beyond individual diagnoses. By aligning with the Research Domain Criteria (RDoC) framework, this approach aims to uncover underlying dimensions of emotional and behavioral dysregulation, helping to improve early screening and support for children with diverse and overlapping mental health needs.

This study evaluated a 90-second movement-based task using a wearable sensor to help identify young children with anxiety or depression. In a sample using machine learning, the approach identified children with internalizing diagnoses at 81% accuracy—outperforming parent-report measures in sensitivity. The findings suggest this method may support earlier, more objective screening of emotional concerns in children.

This editorial outlines the urgent need for objective, scalable mental health assessments and highlights the promise of digital biomarkers captured through wearable devices and smartphones. The authors advocate for investments in multi-modal phenotyping, equitable study design, explainable AI, and user-centered intervention delivery. These strategies could transform mental health screening and treatment by enabling personalized, real-time support—particularly for underserved groups like children and those in rural settings.

This feasibility case study investigates how wearable sensors capturing movement and physiological signals might help characterize children's responses during a brief stress task. Data from two participants suggest certain signals, like sacral movement and heart rate variability, may be useful for understanding individual differences. Findings also highlight opportunities to simplify sensor use in future studies.

In this large study of over 900 families (age 2-6 years), researchers found that most young children who met criteria for a mental health diagnosis were not perceived by their parents as needing help. Parents were more likely to recognize a need when symptoms and impairment were high, especially in cases of depression. Parental mental health also influenced perception—highlighting how both child and parent factors affect whether early concerns are identified and addressed.

This study explored whether voice patterns during a short speech task could help identify signs of anxiety and depression in young children. Using machine learning, researchers analyzed audio features like pitch and repetition, finding that children with internalizing symptoms showed distinct vocal patterns. The approach showed promise as a potential screening aid and outperformed parent-report tools in this sample.

This study surveyed nearly 1,000 U.S. adults to crowdsource real-world, goal-specific wellness practices across five key areas: sleep, productivity, and physical, emotional, and social wellness. Results revealed that preferred wellness strategies were often distinct from research-recommended interventions and varied by health status but not by demographics. A publicly available web dashboard was developed to explore the findings and support future efforts in tailoring sustainable, personalized wellness interventions.

This perspective explores the promises and challenges of using digital phenotype screening tools (DPSTs)—such as wearables—to support mental health screening in children. While these tools may offer more objective and accessible assessments than caregiver reports, the authors highlight critical ethical concerns related to accuracy, privacy, and equity. The paper outlines key principles to guide the responsible development and implementation of DPSTs in pediatric care.

This study examined whether wearable sensors could detect signs of internalizing difficulties—like anxiety or depression—through children’s movement during a joyful, play-based task (blowing bubbles). Researchers found that body motion patterns related to reward responsiveness could help distinguish children with emotional concerns. The results suggest that both positive and negative mood tasks may offer complementary insights for early emotional screening.

This study tested whether a simple belt-worn sensor could capture movement patterns during a fear-based task that relate to anxiety and depression in children ages 3–7. In a sample of 63 children, sensor data were significantly associated with both parent- and clinician-reported internalizing symptoms, while traditional behavioral coding was not. The findings suggest wearable sensors may offer a more feasible and objective way to support early identification efforts.

This study used movement and vocal data from 84 children (ages 4–8) to explore early indicators of anxiety, depression, and ADHD. Machine learning revealed that behavioral patterns were more closely linked to gender and symptom severity than age. Predictive models showed moderate success, supporting the potential of digital tools for early mental health screening.

This commentary expands on McGinnis et al. (2021), advocating for a shift from focusing solely on the child to a broader, family- and relationship-centered approach in early mental health care. It emphasizes the influence of parental perception, cultural context, and the need for transdiagnostic and relational frameworks in recognizing and addressing young children’s emotional needs. The authors call for more inclusive systems that engage all caregivers and improve communication across support networks.