Artificial intelligence in nursing and midwifery: A systematic review

• Published in: Journal of clinical nursing Vol. 32; no. 13-14; pp. 2951 - 2968
• Main Authors: O'Connor, Siobhán, Yan, Yongyang, Thilo, Friederike J. S., Felzmann, Heike, Dowding, Dawn, Lee, Jung Jae
• Format: Journal Article
• Language: English
• Published: England Wiley Subscription Services, Inc 01.07.2023
• Subjects: Adult; Artificial Intelligence; Child; Clinical medicine; Curriculum; deep learning; Delivery of Health Care; Female; healthcare; Humans; machine learning; Midwifery; natural language processing; neural networks; Nursing; Pediatrics; Pregnancy; Professions; deep learning; natural language processing; midwifery; neural networks; nursing; machine learning; artificial intelligence; healthcare
• ISSN: 0962-1067; 1365-2702; 1365-2702
• DOI: 10.1111/jocn.16478

Background Artificial Intelligence (AI) techniques are being applied in nursing and midwifery to improve decision‐making, patient care and service delivery. However, an understanding of the real‐world applications of AI across all domains of both professions is limited. Objectives To synthesise literature on AI in nursing and midwifery. Methods CINAHL, Embase, PubMed and Scopus were searched using relevant terms. Titles, s and full texts were screened against eligibility criteria. Data were extracted, analysed, and findings were presented in a descriptive summary. The PRISMA checklist guided the review conduct and reporting. Results One hundred and forty articles were included. Nurses’ and midwives' involvement in AI varied, with some taking an active role in testing, using or evaluating AI‐based technologies; however, many studies did not include either profession. AI was mainly applied in clinical practice to direct patient care (n = 115, 82.14%), with fewer studies focusing on administration and management (n = 21, 15.00%), or education (n = 4, 2.85%). Benefits reported were primarily potential as most studies trained and tested AI algorithms. Only a handful (n = 8, 7.14%) reported actual benefits when AI techniques were applied in real‐world settings. Risks and limitations included poor quality datasets that could introduce bias, the need for clinical interpretation of AI‐based results, privacy and trust issues, and inadequate AI expertise among the professions. Conclusion Digital health datasets should be put in place to support the testing, use, and evaluation of AI in nursing and midwifery. Curricula need to be developed to educate the professions about AI, so they can lead and participate in these digital initiatives in healthcare. Relevance for clinical practice Adult, paediatric, mental health and learning disability nurses, along with midwives should have a more active role in rigorous, interdisciplinary research evaluating AI‐based technologies in professional practice to determine their clinical efficacy as well as their ethical, legal and social implications in healthcare.