Friday, May 9, 2025
5/9/2025
Artificial Intelligence to Customize Participation-Focused Interventions in Pediatric Re/habilitation - Background. Community and school participation are crucial for the full inclusion of children and youth with disabilities into society. This is also evident in the domain “Community Living and Participation” of the NIDILRR’s research program. There is growing evidence of disparities between children and youth with and without disability in their community and school participation, warranting the need for participation-focused interventions. Two groups with evidence of participation needs include children and youth who receive health-related services and children and youth with craniofacial microsomia (CFM), for which participation needs were newly established. Supporting participation requires creative approaches for customizing interventions to meet the individual needs of families and their children. Artificial Intelligence (AI), including natural language processing (NLP), is often used to tailor information to individuals’ preferences and needs. NLP, therefore, holds promise for supporting customized participation-focused interventions. The Participation and Environment Measure Plus (PEM+) is a technology-based intervention that has high potential to benefit from the use of NLP to customize participation-focused interventions for children and youth. PEM+ was developed with provider and family input. It is designed to guide caregivers online to prioritize participation goals and then use a strategy exchange feature to identify participation-focused strategies for goal attainment. Yet, results of acceptability testing of the PEM+ revealed a need to optimize its strategy exchange feature to provide for more customized user experiences. To address this need, further knowledge in two areas is critically needed: 1) identification of predictors of participation and the specific mediating effect of participation-focused caregiver strategies on participation frequency and involvement; and 2) development of NLP models to classify caregiver strategies according to four known key predictors of participation: environmental support, child and youth’s activity competence, activity preferences, and sense of self. Objectives. This proposed project has two aims: 1) to examine key predictors of community and school participation frequency and involvement for children and youth with disabilities and the indirect effect of participation-focused caregiver strategies on participation; 2) to use NLP to develop a predictive model that classifies participation-focused caregiver strategies according to key predictors of participation. Methods. For both proposed aims secondary analyses of a subset of data from a longitudinal cohort study will be applied. The dataset includes 260 families of children and youth with established participation needs (i.e., 120 children and youth with craniofacial microsomia; heterogenous sample of 140 children and youth receiving services). Because type of disability has been shown to be a weak predictor of participation, prioritization was given to maximizing the sample size in informing the optimization of PEM+. For Aim 1, structural equation modeling (SEM) will be applied to test for the relationship and inter-relationships between potential predictors of participation and participation. For Aim 2, NLP will be used to analyze 1,112 caregiver strategies. Multinomial logistic regression and multinomial naïve Bayes, two supervised machine learning methods, will be applied to develop a predictive model for classifying caregiver strategies to key predictors of participation. Expected Impact. Project results will provide evidence supporting the customized use of participation-focused caregiver strategies to enhance participation goal attainment of children and youth with disabilities. These results, in turn, will advance knowledge about the use of AI to customize electronic participation-focused interventions such as PEM+ to the needs of children and youth with disabilities and support their full inclusion into society.
HHS’ Tracking Accountability in Government Grants System (TAGGS) website provides access to detailed descriptions of grants, loans, aggregated direct payments and other types of financial assistance awarded by the HHS Operating Divisions (OPDIVs) and the Office of the Secretary Staff Divisions (STAFFDIVs).