Leveraging Artificial Intelligence As a Tool for Rendering Family-Centered Services: A Scoping Review

Background and Purpose: Family-centered health work has seen a proliferating use of artificial intelligence (AI) methods, including machine learning, predictive risk modeling, and predictive analysis. However, ethical concerns have also been observed in these services regarding vulnerable populations and family-centered services tailored for such populations. We define family-centered services as those that provide support and resources for the family unit of question, including but not limited to parents, children, adolescents, and couples. Following a database search for studies that employ AI within the context and goal of family-centered services provision, we address the following research questions in this scoping review: (1) How are studies focused on applying AI to render family-centered services collecting family data? (2) How are these studies analyzing family data?; (3) How are these studies/programs providing family-centered services?; (4) Do such studies consider the ethical use of AI? Drawing on interdisciplinary expertise in the fields of psychology, data science, and social work, we evaluate the impact of AI on effective and socially just family-centered services.

Methods: Using the PRISMA Scoping Review (PRISMA Sc-R) framework (Tricco et al., 2018), we analyzed existing literature based on eligibility criteria of: (1) peer-reviewed; (2) written in English articles; (3) published between 2014-2024; (4) involve the use of AI in collecting data, analyzing data, or providing resources in the context of family-centered service provision; (5) have a majority sample that includes couples, parents, children, or adolescents; and (6) be quantitative in analysis. A comprehensive literature search was conducted in these databases: Academic Search Complete; APA PsycInfo; Applied Science & Technology; Family & Society Studies Worldwide; SocINDEX; Communication & Mass Media Complete; PubMed; Excerpta Medica dataBASE (EMBASE); Cochrane Library; Psychology and Behavioral Sciences Collection; and Social Sciences Abstract. Two reviewers independently screened articles, with a third reviewer resolving any screening decision conflicts.

Results: Using multiple databases and a comprehensive search strategy, our search results yielded 6,956 articles spanning the fields of social work, family medicine, health policy and economics, informatics, and more. Maternal and adolescent populations were commonly studied in the identified literature. Studies applied various AI methods to existing empirical data, although the replicability of such techniques for practitioners remains in question. Finally, many articles cited conceptual frameworks involving the ethics of AI and the potential for AI bias, suggesting that multiple fields including social work are thinking through the implications of perpetrating human bias through AI application in family-centered services.

Conclusion and Implications: Understanding the use of AI in providing family-centered services calls upon researchers to investigate how AI is used to serve families, across all domains and in multiple fields of study. Similarly, assessing the ethics of AI use may draw upon both theoretical and rapidly growing empirical research. By evaluating existing analyses on AI use and fairness, we provide informed and evidence-based suggestions and recommendations for future research and practice in this area. We also highlight example articles that have integrated a multidisciplinary perspective or leveraged a multidisciplinary team in implementing their studies.