New independent study results: Babbly's AI algorithm accurately identifies infant babbling patterns
At Babbly, we’ve always emphasized the importance of early detection of language delay and the long term impact it can have on a child’s literacy skills and overall development. Missed babbling milestones are one critical early indicator of potential developmental issues. Infants who experience delays in babbling during their first year of life are more likely to exhibit smaller vocabularies and delayed language development later on. Notably, the absence or significant reduction in babbling can also serve as an early indicator of potential developmental disorders or delays, such as autism spectrum disorder or apraxia. Recognizing these deviations can facilitate early intervention strategies, ultimately improving outcomes for affected children.
To make reliable actionable insights, it’s important to establish how well the model does at predicting different vocalizations in new recordings. We recently collaborated with the Healthcare Innovation and Technology Lab (HITLAB), led by Dr. Stan Kachnowski, Director of Digital Health at Columbia University. The team at HITLAB independently validated our algorithm for the classification of different infant vocalization patterns, and we’re excited to share the results!
Babbly’s algorithm is designed to detect 4 developmentally relevant infant vocalizations (cooing, single syllable babbling, canonical or reduplicated babbling, and variegated babbling). In the present study, its accuracy was evaluated using real-world recordings of infants aged 4-16 months, collected by the HITLAB team. The algorithm's predictions were compared with annotations made by three independent, trained human observers. The evaluation focused on assessing the alignment between the labels provided by annotators and those predicted by the algorithm (see picture below).
Our algorithm demonstrated a high level of accuracy. An F-1 score of 0.91 (or 91% accuracy) indicates that our model is highly accurate and reliable in identifying the target outcomes. Importantly, this accuracy was consistent across infants of different ages and sexes, indicating the algorithm's applicability throughout the stages of preverbal development.
For more information, check out our white paper:
In summary, this external study showcases the potential of Babbly’s AI algorithm for early detection of developmental language delays in infants. By accurately identifying deviations in babbling patterns, we can intervene sooner, and enhance clinicians’ insights by providing them with accurate information otherwise not available in short visits. This research represents a significant step forward in our ongoing commitment to supporting healthy language development in early childhood.
References
McGillion, M., Herbert, J. S., Pine, J., Vihman, M., DePaolis, R., Keren‐Portnoy, T., & Matthews, D. (2017). What paves the way to conventional language? The predictive value of babble, pointing, and socioeconomic status. Child Development, 88(1), 156-166.
Oller, D. K., Eilers, R. E., Neal, A. R., & Schwartz, H. K. (1999). Precursors to speech in infancy: The prediction of speech and language disorders. Journal of Communication Disorders, 32(4), 223-245.
Patten, E., Belardi, K., Baranek, G. T., Watson, L. R., Labban, J. D., & Oller, D. K. (2014). Vocal patterns in infants with autism spectrum disorder: Canonical babbling status and vocalization frequency. Journal of Autism and Developmental Disorders, 44, 2413-2428.
Overby, M., Belardi, K., & Schreiber, J. (2020). A retrospective video analysis of canonical babbling and volubility in infants later diagnosed with childhood apraxia of speech. Clinical Linguistics & Phonetics, 34(7), 634-651.