Session: Technological Advances in Artificial Intelligence and Informatics That Drive Policy and Practice Innovations in Social Work (Society for Social Work and Research 30th Annual Conference Anniversary)

Increasingly, social work researchers are utilizing innovative methodologies and technological advances to drive policy-related advocacy globally and help develop targeted interventions with specialized populations who have the greatest need. Machine learning (ML) is a powerful tool to enhance human performance and transform the delivery of care of individuals and the communities social workers serve, which can be applied to a number of emerging and increasingly pressing social and behavioral health concerns, including gambling disorder. Additionally, Geographic Information Systems (GIS) mapping allows social workers to explore socioeconomic and demographic variables by location and visualize data to guide the development of region-specific services and mental health workforce development. Finally, in regions of the world where data collection infrastructures are underdeveloped, language models and social media data can help build information pathways and novel pipelines to enhance development efforts.

In this symposium, participants will discuss artificial intelligence applications and advances in data science, which are utilized to assist with workforce, policy, practice, and global development efforts and identify targeted, accessible, and culturally appropriate services. In the first paper, machine learning models are used to predict problem gambling risk among online casino bettors to inform targeted interventions. Highly correlated features (|r| > 0.9) were excluded, and dimensionality was reduced via principal component analysis. The model achieved a strong silhouette score of 0.86, indicating high intra-cluster similarity and strong inter-cluster separation among three distinct groups. In the second paper, innovative GIS mapping provides geographically based depictions of those who are most likely to experience gambling-related harms. Harm risk levels by zip code were calculated as weighted composite scores, based on the strength of indicators (i.e. point-biserial correlations) related to problem gambling in a statewide prevalence survey. In the third paper, using the case of Zambia, a language model, exploring Twitter discourse as a proxy for poverty metrics, captures salient development issues across regions and time. Parsimonious, interpretable topic-based features accounted for more than 60\% of the variation in village-level wealth. Results showed that the spatial interpolation method, kriging, outperformed the commonly used imputation methods. Participants will discuss the appropriate and ethical use of AI, ML, and GIS with specialized populations globally.