Leveraging Artificial Intelligence to Inform Care Coordination by Identifying and Intervening in Patients' Unmet Social Needs: A Scoping Review.

Aim: We reviewed how artificial intelligence has been applied to inform care coordination by identifying and/or intervening in patients' unmet social needs.

Design: Scoping review.

Data Sources: PubMed, CINAHL, PsycInfo, and Scopus databases were searched for articles published by November 2023.

Methods: Articles were excluded if they were reviews or protocols, did not explicitly mention artificial intelligence, or did not primarily focus on using it to identify and/or address unmet needs to inform care coordination.

Results: Of 476 articles that underwent title and abstract screening, 102 were assessed for full-text eligibility, and eight were ultimately included. Five articles used both natural language processing and machine learning; two articles used natural language processing; and one article used machine learning. Half (n = 4) of the articles focused on using artificial intelligence to identify/predict social needs, and two each focused on artificial intelligence to examine social resource provision or to indirectly identify social needs or using artificial intelligence to facilitate addressing unmet needs through care coordination.

Conclusions: This review can inform an understanding of facilitators and barriers to the implementation of artificial intelligence in practice, to potentially improve patient care, health outcomes, and population health equity.

Implications For Patients And The Profession: Using artificial intelligence to promote care coordination can expand opportunities to identify and intervene on social needs across more patients, with implications for nurses and other health professionals. It can also potentially exacerbate inequities and harm patient trust.

Impact: The findings suggest a gap between the practice of incorporating artificial intelligence into integrated care platforms and the available scientific literature. This review can provide healthcare providers and organisations with insights into integrating artificial intelligence into clinical workflows, which may inform decisions about whether or how to implement these technologies in clinical settings.

Reporting Method: We followed PRISMA-ScR guidelines. No Patient or Public Contribution.