Digital Phenotyping To Screen For Early Childhood Internalizing Disorders

点击翻译按钮获取中文摘要

ABSTRACT Childhood anxiety and depression are common, impairing, and have the potential to disrupt development well into adulthood. Pediatricians need mental health screening tools to meet the prevalence of these internalizing disorders early in childhood. We have developed a promising digital health approach for detecting movement and speech phenotypes of internalizing disorders in young children. My long-term goal is to create a novel technology for screening young children for internalizing disorders at scale. My short term objective is to leverage my original approach and my proposed training curriculum to create a prototype mobile health (mHealth) application that provides a composite digital phenotype to detect childhood internalizing disorders. My specific aims are to (1) evaluate the validity of digital phenotypes measured by a novel mHealth system for identifying internalizing disorders in young children and (2) develop a composite digital phenotype for childhood internalizing disorders involving multiple constructs from the Research Domain Criteria (RDoC) and identify moderating RDoC construct variables. If successful, these mHealth enabled digital phenotypes may be used to further NIMH health initiatives of enabling better tracking of changes in internalizing mental health status across childhood and supporting new and innovative research-practice partnerships with pediatrics to improve dissemination of evidence-based mental health screening. Similarly, this approach answers a need identified by the American Academic of Pediatrics for new tools that screen for behavioral and emotional problems. Seventy children between the ages of 4 and 8 years, oversampled for internalizing disorders, will be recruited for this study from pediatric and childhood mental health services at a university-based regional medical center that services the state of Vermont and upstate New York. They will be administered mood induction tasks design to press for RDoC Negative and Positive Valence constructs, while instrumented with a belt-worn smartphone enabled with the prototype mHealth application. A composite digital phenotype will be developed across constructs based on statistical classification models trained using machine learning on data captured during the tasks. To carry out this work as an independent investigator, I propose an intensive training curriculum to gain foundational skills in digital phenotyping. It includes training in (1) developmental epidemiology to better research early childhood psychopathology through a public health lens, and (2a) mobile app development to communicate effectively with app developers (2b) analysis of complex systems including machine learning approaches, (2c) ethical and societal issues regarding digital psychiatry. Completion of these training and research aims will provide me the skills and evidence to develop an easily administered digital health technology for identifying young children with internalizing disorders in the short-term, and in the long term, the ability to successfully contribute to the emerging field of digital health for the purpose of improving childhood mental health.