Rai, Richa. New techniques to facilitate longitudinal video-based digital tracking of infant development. Retrieved from https://doi.org/doi:10.7282/t3-6d0p-t048
DescriptionIn the last few decades, the rate of neurodevelopmental disorders has increased. However, current criteria to detect developmental differences rely on social and emotional parameters determined by observation alone. This reliance on observation restricts the description of movements to those that are less ambiguous, thus leaving out important components of the continuous stream of motions that make up natural behaviors. Movements that occur spontaneously and lend fluidity to social behavior occur largely beneath awareness and escape the naked eye of the observer. Furthermore, they are present since birth, unlike the more reliable social movements that appear later in life, when an infant’s motor system has matured. We here take the approach of examining movements since birth and tracking their maturation, to determine natural ranges of typical neurodevelopment. We leverage off-the-shelf technology and new advances in computer vision to bring the parents into the research loop, empowering them as active contributors to cognitive developmental research. Two families submitted weekly videos of their infants starting at the age of 10 days, and continuing for 17 weeks, both infants comprise a longitudinal data set tracking the infants over the first 17 weeks. In addition, videos of infants were acquired from open access sources, which when combined with the longitudinal set, comprise a cross-sectional dataset. The videos were processed through OpenPose, and using these videos, we develop proof of concept that simple video-based assessment is possible from the comfort of the home. This new model can provide both longitudinal and cross-sectional data, to shed light on the evolution, since birth, of basic building blocks that will make up social motor behavior, possibly including atypical patterns detectable years later. Our work introduces new methods of data acquisition and analysis that can advance the study of neuromotor control in neonates and provide the means for much earlier detection of neurodevelopmental derail.