Objective child behavior measurement with naturalistic daylong audio recording and its application to autism identification.
ABSTRACT Child behavior in the natural environment is a subject that is relevant for many areas of social science and bio-behavioral research. However, its measurement is currently based mainly on subjective approaches such as parent questionnaires or clinical observation. This study demonstrates an objective and unobtrusive child vocal behavior measurement and monitoring approach using daylong audio recordings of children in the natural home environment. Our previous research has shown significant performance in childhood autism identification. However, there remains the question of why it works. In the previous study, the focus was more on the overall performance and data-driven modeling without regard to the meaning of underlying features. Even if a high risk of autism is predicted, specific information about child behavior that could contribute to the automated categorization was not further explored. This study attempts to clarify this issue by exploring the details of underlying features and uncovering additional behavioral information buried within the audio streams. It was found that much child vocal behavior can be measured automatically by applying signal processing and pattern recognition technologies to daylong audio recordings. By combining many such features, the model achieves an overall autism identification accuracy of 94% (N=226). Similar to many emerging non-invasive and telemonitoring technologies in health care, this approach is believed to have great potential in child development research, clinical practice and parenting.
SourceAvailable from: Umit Yapanel[Show abstract] [Hide abstract]
ABSTRACT: Speech signal processing and other man-machine interaction technologies have been developed for improved child-computer interaction for education, entertainment, as well as other applications (1, 2). However, for very young children (in the age range of 0 to 4 years old, and especially 0 to 2), such interaction is not encouraged (3, 4). Instead, parent-child interaction is highly recommended (3, 4) since it promotes improved language development. In this study, a new system entitled LENA TM (Language Environment Analysis) and its associate processing technologies will be introduced. LENA provides parents/caregivers with quantified statistical information concerning the language environment and development status of children in order to allow for the determination of what needs to improve and how to improve. The adult word count (AWC) estimation algorithm is shown to reduce the relative Root Mean Square Error from an initial 42% to 7-8% after 5 hours of measuring time. If LENA's feedback suggests any potential development problem, parents can take action at a crucial early stage. LENA is a new processing system not only for parents/caregivers but for pediatricians, speech language pathologists, child development psychologists, and other researchers as well. This system represents one of the first breakthroughs in assessing early childhood language development and child environment conditions.
[Show abstract] [Hide abstract]
ABSTRACT: Early identification is crucial for young children with autism to access early intervention. The existing screens require either a parent-report questionnaire and/or direct observation by a trained practitioner. Although an automatic tool would benefit parents, clinicians and children, there is no automatic screening tool in clinical use. This study reports a fully automatic mechanism for autism detection/screening for young children. This is a direct extension of the LENA (Language ENvironment Analysis) system, which utilizes speech signal processing technology to analyze and monitor a child's natural language environment and the vocalizations/speech of the child. It is discovered that child vocalization composition contains rich discriminant information for autism detection. By applying pattern recognition and machine learning approaches to child vocalization composition data, accuracy rates of 85% to 90% in cross-validation tests for autism detection have been achieved at the equal-error-rate (EER) point on a data set with 34 children with autism, 30 language delayed children and 76 typically developing children. Due to its easy and automatic procedure, it is believed that this new tool can serve a significant role in childhood autism screening, especially in regards to population-based or universal screening.Conference proceedings: ... Annual International Conference of the IEEE Engineering in Medicine and Biology Society. IEEE Engineering in Medicine and Biology Society. Conference 01/2009; 2009:2518-22. DOI:10.1109/IEMBS.2009.5334846
[Show abstract] [Hide abstract]
ABSTRACT: The study compared the vocal production and language learning environments of 26 young children with autism spectrum disorder (ASD) to 78 typically developing children using measures derived from automated vocal analysis. A digital language processor and audio-processing algorithms measured the amount of adult words to children and the amount of vocalizations they produced during 12-h recording periods in their natural environments. The results indicated significant differences between typically developing children and children with ASD in the characteristics of conversations, the number of conversational turns, and in child vocalizations that correlated with parent measures of various child characteristics. Automated measurement of the language learning environment of young children with ASD reveals important differences from the environments experienced by typically developing children.Journal of Autism and Developmental Disorders 11/2009; 40(5):555-69. DOI:10.1007/s10803-009-0902-5 · 3.06 Impact Factor