Challenges of Artificial Intelligence Applications in Healthcare Sector: A Scoping Review

Abstract

Introduction: The field of Artificial Intelligence (AI) is seeing tremendous growth and has significant promise for enhancing service delivery. The emergence of AI has resulted in a fundamental transformation in multiple industries, notably the healthcare sector. Artificial Intelligence (AI), due to its capacity to replicate human cognitive processes and acquire knowledge through experiential learning, has the capability to significantly transform healthcare services. Nevertheless, notwithstanding the prospective advantages, there exists several obstacles that necessitate resolution to comprehensively incorporate and optimize the capabilities of AI inside the healthcare domain.

Objectives: This scoping review aims to explore the challenges of AI applications in healthcare settings.

Methods: Database search was conducted involving five databases namely Science Direct, Scopus, PubMed, ProQuest, and Google Scholar using Boolean Operators for relevant terms. Included studies involved studies on different healthcare settings and focus on AI and its challenges in terms of implementation. Non-original articles and articles which did not fit into the research questions and were not aligned with AI or machine learning definitions were excluded.

Results: A total of 2794 articles were published between January 2018 to September 2023, with 21 articles included in this review. Most of the studies were conducted in North America (38.1%, n = 8), followed by Europe (28.5%, n = 6), Asia (19.0%, n = 4), Worldwide (9.5%, n = 2), and Africa (4.7%, n = 1). Most of the studies were cross-sectional (42.8%, n= 9). The most common healthcare setting was acute care (64.5%, n = 60), followed by qualitative study (19.0%, n=4), predictive study (14.2%, n= 3), case-control study (9.5%, n=2) and one each of mixed method and randomized control trial. Key challenges identified across studies encompassed workforce readiness and training gaps, limited transparency and interpretability of AI systems, data quality and bias concerns, infrastructural and resource limitations, ethical and legal uncertainties, risks to patient–clinician relationships, and potential exacerbation of health inequities.

Conclusions: This review's findings emphasize the need for a well-considered and holistic approach to integrating AI into healthcare to maximize its benefits while addressing difficulties and guaranteeing patient safety and well-being.