Implementation of artificial intelligence in healthcare: a scoping review

Article type
Authors
Tricco AC1, Darvesh N2, Thomas SM2, Fennelly O3, Brar R2, Straus SE4
1Li Ka Shing Knowledge Institute, St. Michael’s Hospital, Unity Health Toronto; Epidemiology Division, Dalla Lana School of Public Health, University of Toronto
2Li Ka Shing Knowledge Institute, St. Michael’s Hospital, Unity Health Toronto
3Irish Centre for High End Computing (ICHEC), National University of Ireland (NUI) Galway
4Li Ka Shing Knowledge Institute, St. Michael’s Hospital, Unity Health Toronto; Department of Medicine, University of Toronto
Abstract
Background:
Artificial intelligence (AI) has the potential to lead to significant improvements in health care and public health. This includes using data analytics to personalize health care for patients, assist health care professionals, and tailor organizational and policy decision-making. However, to optimize the implementation of AI across health care in a safe, effective and sustainable manner, the current implementation strategies and outcomes at both the patient and population-level need to be examined.

Objectives:
To conduct a scoping review to identify what strategies are used to implement AI interventions for health or within healthcare systems.

Methods:
The Joanna Briggs Institute reviewer’s manual will guide the conduct of this review and the review protocol has been published. The eligibility criteria for this review includes:

- Population: Adults and children of any age.
- Intervention: Implementation of AI tools for health or within a health system.
- Comparator: Any.
- Outcome: Any outcome at the patient, public, clinician, population or system level.
- Study Design: All primary experimental, cohort and case-control studies.
- Year published: Limited to 2008 onwards.

A search strategy will be developed by an experienced information specialist and peer-reviewed. Multiple databases will be searched as well as grey literature and reference lists of included studies. Identified articles will be screened by titles and abstract and then by full text by pairs of reviewers, with discrepancies resolved by a third reviewer. A standardized charting form will be used to extract data from the included studies by pairs of reviewers independently, with study authors contacted where information is missing or unclear.

Results:
Results will include the types of AI tools implemented for health or within a health system, the implementation strategies, the participants, and outcomes which may include sustainability, scalability, barriers and facilitators.

Conclusions:
The findings from this review will identify areas where AI has been implemented in health care, how it has been implemented, and the outcomes at patient, public, clinician, population, and system levels. This will inform future research on AI implementation. In the long term, it will also help inform strategies for the safe, effective, and sustainable implementation of AI tools in health care in order to improve health care quality.

Patient or healthcare consumer involvement: not applicable