Icon Legend

This session is not in your schedule.

This session is in your schedule. Click again to remove it.

Presentation Icons

Recorded Session

Ticketed Session

333 Views

ViewAttendees 1

Symposia

Technology/Digital Health

Leveraging Novel Scientific Approaches and Digital Technologies to Bridge the Divide in Under Addressed Psychological Conditions

2 - (SYM 25) Concurrent, Predictive, and Incremental Validity of Natural, Daylong Cry Audio Recordings and a Machine Learning Algorithm to Identify Risk for Irritability and Psychopathology in Young Children

Friday, November 21, 2025

3:27 PM - 3:48 PM CST

Location: Bolden 2, Level 2

Keywords: Technology / Mobile Health, Vulnerable Populations, Anger / Irritability
Recommended Readings: Philippe, T. J., Sikder, N., Jackson, A., Koblanski, M. E., Liow, E., Pilarinos, A., & Vasarhelyi, K. (2022). Digital health interventions for delivery of mental health care: systematic and comprehensive meta-review. JMIR mental health, 9(5), e35159., Chung, J., & Teo, J. (2022). Mental health prediction using machine learning: taxonomy, applications, and challenges. Applied Computational Intelligence and Soft Computing, 2022(1), 9970363., van Genugten, C. R., Thong, M. S., Van Ballegooijen, W., Kleiboer, A. M., Spruijt-Metz, D., Smit, A. C., ... & Riper, H. (2025). Beyond the current state of just-in-time adaptive interventions in mental health: a qualitative systematic review. Frontiers in Digital Health, 7, 1460167., Larkin, M., Boden, Z. V., & Newton, E. (2015). On the brink of genuinely collaborative care: experience-based co-design in mental health. Qualitative health research, 25(11), 1463-1476.,

Speaker(s)

Lauren M. Henry, Ph.D. (she/her/hers)

Postdoctoral Fellow
National Institute of Mental Health
Bethesda, MD, United States

Co-author(s)

KL

Kyunghun Lee, Ph.D.

Staff Scientist
National Institute of Mental Health
Bethesda, MD, United States
ET

Elizabeth Tandilashvili, BS

Postbaccalaureate Intramural Research Training Award Fellow
National Institute of Mental Health
Bethesda, MD, United States
EH

Ellie Hansen, BA

Postbaccalaureate Intramural Research Training Award Fellow
National Institute of Mental Health
Bethesda, MD, United States
FP

Francisco Pereira, PhD

Director, Machine Learning Team
National Institute of Mental Health
Bethesda, MD, United States
Daniel S. Pine, M.D.

Chief, Section on Development and Affective Neuroscience
National Institute of Mental Health
Bethesda, MD, United States
EN

Elizabeth Norton, D. Phil.

Associate Professor, Department of Communication Sciences and Disorders
Northwestern University
Evanston, IL, United States
LW

Lauren Wakschlag, PhD

Director, Professor
Northwestern University
Chicago, IL, United States
MB

Melissa Brotman, PhD

Principal Investigator
Emotion and Development Branch, NIMH
Bethesda, MD, United States

Abstract Body

Background: There are individual differences in cry - a common, emotionally salient vocalization and primary way infants/young children communicate needs to caregivers. Machine learning can be leveraged to automatically detect cry in natural, daylong audio recordings, facilitating the study of cry as a potential early, ecologically valid, and scalable indicator of irritability and later psychopathology risk. Here, we examine cry’s (1) concurrent validity with irritability in young children; (2) predictive validity with later psychopathology; and (3) incremental validity relative to irritability measures in predicting later psychopathology.

Methods: 356 toddlers oversampled for high irritability (Mage=14.7 mos, 45.8% female) and their biological mothers (Mage=31.4 years, 50.6% White, 59.8% household income < $100,000) participated in the [BLIND] study at [BLIND]. Toddlers completed natural, daylong audio recordings via the Language ENvironment Analysis system. We developed and evaluated a binary classifier (i.e., combining traditional and deep spectrum audio features as input to a Gradient Boosting Machine classifier) to identify cry in each audio recording. The area under the receiver operating characteristic curve of the classifier (relative to ground truth hand annotations) in the current dataset is 0.902. We measured irritability via parent report (Multidimensional Profile of Disruptive Behavior-Infant-Toddler) and a standardized lab task (Disruptive Behavior Diagnostic Observation Schedule). One year later, parents completed structured interviews of child psychopathology (Preschool Age Psychiatric Assessment).
Results: Aim 1: Greater duration of cry was associated with more parent-reported irritability (β=.16, F(1,196)=5.39, p=.02, adj R²=.02) and unrelated to lab-based irritability (β=-.02, F(1,243)=.13, p=.72, adj R²=.001). Aim 2: Greater duration of cry predicted more externalizing symptoms one year later (β=.33, F(1,150)=17.94, p< .001, adj R²=.10). The association between greater cry duration and more internalizing symptoms approached significance (β=.16, F(1,150)=4.00, p=.05, adj R²=.02). Aim 3: Cry (β=.18, p=.03) contributed unique information in the prediction of externalizing symptoms one year later, relative to parent report (β=.46, p< .001) and a lab-based task of irritability (β=.08, p=.36), F(3,113)=16.39, p< .001, adj R²=.29.

Conclusions: Cry may be promising as part of a scalable, early risk assessment toward evidence-based preventive treatment for psychopathology.