Public and Patient Perspectives on AI in Health Information and Decision Making: A Scoping Review Protocol

Sarah Elliott; Bani Sekhon; Samantha Guitard; Lisa Hartling

doi:10.20944/preprints202606.1993.v1

Submitted:

25 June 2026

Posted:

26 June 2026

You are already at the latest version

Abstract

Objective: The objective of this scoping review is to identify and map the extent, range, and nature of the literature regarding the attitudes, perceptions, opinions, and trust levels of patients and the general public when using Artificial Intelligence (AI) for health information seeking and decision making.Introduction: AI is rapidly transforming healthcare by providing tools that enhance diagnostic accuracy, streamline clinical workflows, personalize treatment recommendations, and empower patients with more effective self-management strategies. However, adoption depends on public acceptance, which is often hindered by a "trust-preference paradox" - where users may trust AI’s technical capabilities but prefer human oversight for final decisions. This review is necessary to clarify these evolving concepts and identify research gaps in non-clinical AI usage for health information.Inclusion Criteria: Following the PCC (Population, Concept, and Context) framework, this review will include laypeople and patients (Population) regarding their attitudes, perceptions, opinions and trust in AI tools used for health information seeking and self-management (Concept) outside professional clinical settings (Context).Methods: A three-step search strategy will be utilized across databases including MEDLINE, Embase, CINAHL, PsycINFO, Web of Science and Semantic Scholar. Two independent reviewers will perform study selection and data extraction using a standardized charting form. Results will be presented through tabular summaries and narrative synthesis.

Keywords:

public and patient perspectives

;

artificial intelligence

;

health information

;

scoping review

Subject:

Public Health and Healthcare - Public Health and Health Services

Introduction

Artificial Intelligence (AI) was first defined by John McCarthy in 1956 as the “science and engineering of making intelligent machines” designed to simulate human cognitive functions such as learning, reasoning, and problem-solving (Zondag et al., 2024). After decades of fluctuating interest, AI has seen a major resurgence since the 1990s, reflected in the recent popularity of large language models (LLMs) like ChatGPT (Chen et al., 2024). In the healthcare sector, AI is no longer a distant theoretical promise; it is currently being integrated into many aspects of care such as screening, diagnosis, and treatment planning, with global expenditure on these technologies projected to grow exponentially (Chen et al., 2024; Zondag et al., 2024).

Current evidence suggests that AI has the potential to improve multiple aspects of healthcare by increasing efficiencies and processing complex, high-dimensional datasets to support more personalized health recommendations. However, the successful integration of AI is not merely a technical challenge of “recall and precision” but a socio-technical one that depends heavily on public acceptance and user trust. Emerging research indicates a complex landscape of public attitudes towards the use of AI in healthcare (Witkowski et al., 2024; Chew et al., 2024). For instance, some reports suggest that patients may be “more honest” with AI than with human doctors regarding sensitive health information (Cao et al., 2026). Conversely, other studies highlight significant “patient apprehensions” regarding the potential erosion of the “human touch,” algorithmic bias, and the loss of patient-driven choice in care (Maleki et al., 2026). Concerns have also been raised about the growing use of AI for health information seeking, including the accuracy, reliability, transparency, and trustworthiness of AI-generated information. As increasing numbers of individuals turn to AI tools to obtain health information and inform personal health decisions, questions remain regarding how users evaluate AI-generated content, the extent to which they rely on it when making health-related decisions, and the potential risks associated with misinformation, incomplete information, or overreliance on AI recommendations.

Rationale

The rationale for this review stems from the fragmented nature of the existing literature on how members of the public interact with AI outside of clinical settings (Aras et al., 2026). While significant research has focused on AI-based clinical decision support systems within hospital workflows (Zondag et al., 2024), there is less clarity regarding self-directed usage. Patients are increasingly using AI as a “longitudinal boundary companion” that scaffolds their health journey from initial symptom appraisal to post-consultation sense-making (Maleki et al., 2026). Despite this, existing reviews often focus on specific sub-populations (e.g., US-only samples) or specific health professional perspectives, leaving a gap in our understanding of the broader public’s experience with AI for health information seeking.

A scoping review is the most appropriate methodology for this inquiry because the research question is broad and exploratory in nature. Unlike a systematic review, which typically seeks to determine the effectiveness of a specific intervention, a scoping review allows for the identification and mapping of the breadth of available evidence, the clarification of evolving concepts like “AI trust,” and the identification of research gaps and opportunities. This approach is essential for emerging fields where evidence is widely dispersed across different research designs and disciplines.

Operational Definitions

To ensure conceptual clarity, this review will utilize the following operational definitions:

Artificial Intelligence (AI): Advanced technologies that enables machines to effectively complete tasks that typically require humans due to their high complexity (Thomas et al., 2026). Typical examples include symptom checkers (e.g., Ada Health), AI-powered health platforms (e.g., Google Health), and generalized LLMs (e.g., ChatGPT).
AI application: Software programs that perform either simple or complex tasks with AI techniques (GoogleCloud, n.d.).
AI system: Machines that can learn from large amounts of data and examples, to identify patterns in information and make decisions or predictions without explicit programming for each possible scenario (GoogleCloud, n.d.)
Trust: A user’s belief in the AI technology’s reliability, accuracy, and fairness (Valenzuela et al., 2025).
Attitudes and Perceptions: The subjective opinions, feelings of “excitement or concern,” and perceived usefulness of AI tools for health information (Maleki et al., 2026).
Health Information Seeking: The pursuit of “personalized and actionable” medical knowledge to understand health status, symptoms, or treatment options (Wu et al., 2023). This involves the direct use of AI to help laypeople articulate health concerns and interpret clinical advice across their healthcare-seeking trajectory (Cao et al., 2026).
Decision Making: The process of “negotiating healthcare choices,” where users analyze health signals and compare potential treatment pathways to actively participate in their own care. This role empowers patients to transition from passive recipients to active negotiators capable of challenging or collaboratively refining treatment plans with clinicians (Cao et al., 2026).

Preliminary Search

A preliminary search of MEDLINE, the Cochrane Database of Systematic Reviews, and JBI Evidence Synthesis was conducted. While existing systematic reviews (e.g., Beets, 2023; Wu et al., 2022) address public perceptions, they are often limited by geographic scope or focus specifically on clinical applications. This proposed scoping review will differ by mapping the broader public’s self-directed use of AI in non-clinical contexts, providing a map of the socio-technical landscape outside the clinical setting.

Review Objective

The objective of this scoping review is to identify and map the extent, range, and nature of the literature regarding the attitudes, perceptions, opinions, and trust levels of patients and the general public when using AI for health information seeking and decision making outside of a clinical setting.

Review Questions

What is the extent and nature of the scientific literature about patients’ and the general public’s attitudes, perceptions, opinions of, and trust with, using AI for health information seeking and decision making?
How and among whom has this concept been evaluated and what data collection methods and tools have been used?

Inclusion Criteria

Participants

This review will consider studies involving patients or the general public (laypeople) of any age from any country. Studies focusing exclusively on healthcare professionals, AI developers, or students within a training context will be excluded. Studies with overlapping populations will only be considered if data are available separately for the patient/public group.

Concept

The review will include studies examining attitudes, perspectives, opinions, and trust associated with use of AI systems/applications for health information seeking and decision making.

Context

The context is health information seeking and decision making outside of a clinical setting using an AI system/application (e.g., ChatGPT, Claude, and AI chatbots that are publicly available). This includes experiences with general health information seeking, symptom checking, self-diagnosis, comparing treatment options, support with acute and chronic conditions, and when to seek health services.

Types of Sources

This review will consider primary research of any design (e.g., qualitative, quantitative, mixed-methods). It will also include grey literature, e.g., reports and theses. Conference abstracts will not be included, but will be used to search for associated full reports/publications. Editorials, commentaries, and opinion pieces will be excluded.

Methods

The proposed scoping review will be conducted in accordance with the JBI methodology for scoping reviews (Aromataris & Munn, 2020). The final review will be reported following the Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) statement (Tricco et al, 2018).

Search Strategy

The search strategy will aim to locate both published and unpublished studies. A three-step search strategy will be utilized in this review. First, an initial limited search of MEDLINE (PubMed) and CINAHL (EBSCO) was undertaken to identify articles on the topic. The text words contained in the titles and abstracts of relevant articles, and the index terms used to describe the articles, were used to develop a full search strategy for MEDLINE, Embase, PsycINFO, CINAHL, Web of Science, and Semantic Scholar (see Appendix I). The search strategy, including all identified keywords and index terms, was adapted for each included database and information source. The search also covered grey literature sources, including theses, dissertations, and conference abstracts; preprints will be excluded.

Study/Source of Evidence Selection

Following the search, all identified citations will be collated and uploaded to Covidence where duplicates will be removed, and then uploaded into the JBI System for the Unified Management, Assessment and Review of Information (JBI SUMARI) (JBI, Adelaide, Australia; Munn et al., 2019) for screening. Pilot testing will involve screening 100 titles and abstracts by review team members and discussing discrepancies. Additional sets of 100 citations will be screened until reviewers reach adequate agreement. Inclusion and exclusion criteria will be refined after pilot testing to ensure clarity (Table 1). Following the pilot testing, each title and abstract will then be screened by two independent reviewers for assessment against the inclusion criteria. Potentially relevant sources will be retrieved in full and their citation details imported into JBI SUMARI (JBI, Adelaide, Australia). Pilot testing will involve screening 25 full texts by review team members and discussing discrepancies. Additional sets of 25 full texts will be screened until reviewers reach adequate agreement. Subsequently, all full texts will be assessed in detail against the inclusion criteria by two independent reviewers. Reasons for exclusion of full text articles that do not meet the inclusion criteria will be recorded and reported in the scoping review. Any disagreements that arise between the reviewers at each stage of the selection process will be resolved through discussion, or with an additional reviewer. The reference lists of all included studies will be screened for additional studies. References from any relevant reviews (e.g., systematic review, scoping review) identified during screening will also be checked for additional studies. Studies published in any language will be eligible for inclusion. The results of the search and the study inclusion process will be reported in full in the final scoping review and presented in a PRISMA flow diagram.

Data Extraction

Data will be extracted from papers included in the scoping review by one reviewer using a data extraction tool developed by the reviewers, with a second reviewer verifying the extracted data. The data extracted will include details about the participants (patients and general public), concept (attitudes, perceptions, opinions, and trust; type of AI and details of any specific AI tools), context (type of health information, types of decisions; geographic location/scope), study design and methods (e.g., data collection instruments used, concepts examined and definitions of these), and key findings relevant to the review questions. The draft data extraction tool will be pilot-tested on a subset of three included studies and will be modified and revised as necessary during the extraction process. Modifications will be detailed in the final scoping review. Any disagreements between the reviewers will be resolved through discussion or an additional reviewer. If appropriate, authors of papers will be contacted by email up to two times (two weeks apart) to request missing or additional data.

Data Analysis and Presentation

The information presented will directly respond to the review objective and question regarding the extent and nature of the existing literature about public attitudes, perspectives, opinions, and trust. The data will be presented in tabular, graphical and diagrammatic form to illustrate the extent and nature of the literature. This mapping will describe the types of studies (e.g., study designs, data collection methods), demographic characteristics (e.g., population, geographic region), AI application type (e.g., LLMs, chatbots, symptom checkers), and key findings. A narrative summary will accompany the charted results, describing how the findings relate to the review’s objective of understanding the nature of existing research on the socio-technical landscape of AI in personal health management, as well as research gaps.

Knowledge User Engagement

Knowledge user engagement will be integrated throughout this scoping review. We will involve members of our established Patient and Public Advisory Group (P-PAG) to seek their input on key processes to ensure the review remains relevant to the needs and priorities of patients and the public. One member of the P-PAG will serve as a knowledge user on the review team and will contribute to the refinement of the review objectives, interpretation of findings, and planning of knowledge mobilization activities. Additional members of the P-PAG will be consulted at key stages of the review to provide feedback on the relevance, clarity, and applicability of the findings. Engaging knowledge users throughout the review process will help ensure that the review findings are meaningful, patient-oriented, and useful for informing future research, practice, and dissemination activities related to language access and health communication.

Supplementary Materials

The following supporting information can be downloaded at the website of this paper posted on Preprints.org.

Author Contributions

BS: Conceptualization, literature searching, and preparation of the initial draft. SG: Methodology and manuscript review and editing. SE and LH: Conceptualization, methodology, manuscript review and editing, and supervision.

Funding

No funding was received for the conduct of this review.

Acknowledgments

The authors acknowledge Liz Dennett, Research Librarian at the JBI University of Alberta Centre for Knowledge Mobilization, for her assistance in developing the search strategy.

Conflicts of Interest

The authors declare that they have no conflicts of interest.

Declarations

Artificial intelligence tools were not used in the preparation of this protocol.

References

Aras, S., Drakos, C., Vineesha Manimangalam, Nasir, M. A., Burns, C., Smith, D., & Ozlem Equils. (2026). Influencing public acceptance of artificial intelligence (AI) in healthcare delivery. Frontiers in Digital Health, 7. [CrossRef]
Aromataris, E., Munn, Z. (Eds.). (2020). JBI manual for evidence synthesis. JBI. [Accessed June 23, 2026]. Available from: https://synthesismanual.jbi.global. [CrossRef]
Beets, B., Newman, T. P., Howell, E. L., Bao, L., & Yang, S. (2023). Surveying Public Perceptions of Artificial Intelligence in Health Care in the United States: Systematic Review. Journal of medical Internet research, 25, e40337. [CrossRef]
Cao, Y., Ji, Y., Fu, Y., Dharmavaram, S., Turchioe, M., Benda, N. C., Mamykina, L., Sun, Y., & Xu, X. (2026). More than decision support: Exploring patients’ longitudinal usage of large language models in real-world healthcare-seeking journeys. Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems, 1–24. [CrossRef]
Chen, S.-Y., Kuo, H. Y., & Chang, S.-H. (2024). Perceptions of ChatGPT in healthcare: Usefulness, trust, and risk. Frontiers in Public Health, 12. [CrossRef]
Chew, H. S. J., & Achananuparp, P. (2022). Perceptions and needs of artificial intelligence in health care to increase adoption: Scoping review. Journal of Medical Internet Research, 24(1), e32939. [CrossRef]
Google Cloud. What are AI applications? [Accessed June 23, 2026]. Available from: https://cloud.google.com/discover/ai-applications.
Google Cloud. Artificial intelligence (AI): a simple-to-understand guide. [Accessed June 23, 2026]. Available from: https://cloud.google.com/learn/what-is-artificial-intelligence.
Maleki, M. S., Sahebi, L., Shahvari, Z., & Samadi, S. (2026). Attitudes of the Iranian public toward the clinical use of artificial intelligence in medicine: A cross-sectional survey. Patient Preference and Adherence, Volume 20, 1–10. [CrossRef]
Munn, Z., Aromataris, E., Tufanaru, C., Stern, C., Porritt, K., Farrow, J.R., Lockwood, C.S., Stephenson, M.D., Moola, S., Lizarondo, L., McArthur, A., Peters, M.D., Pearson, A.S., & Jordan, Z. (2019). The development of software to support multiple systematic review types: the Joanna Briggs Institute System for the Unified Management, Assessment and Review of Information (JBI SUMARI). International Journal of Evidence-Based Healthcare, 17, 36–43. [CrossRef]
Tricco, A. C., Lillie, E., Zarin, W., O’Brien, K. K., Colquhoun, H., Levac, D., Moher, D., Peters, M. D. J., Horsley, T., Weeks, L., Hempel, S., Akl, E. A., Chang, C., McGowan, J., Stewart, L., Hartling, L., Aldcroft, A., Wilson, M. G., Garritty, C., Lewin, S., … Straus, S. E. (2018). PRISMA Extension for Scoping Reviews (PRISMA-ScR): Checklist and Explanation. Annals of internal medicine, 169(7), 467–473. [CrossRef]
Thomas, J., Flemyng, E., Noel-Storr, A. (updated March 2026). Responsible use of AI in evidence SynthEsis (RAISE): building and evaluating AI evidence synthesis tools. In: Open Science Framework Washington, DC: Center for Open Science [Accessed June 23, 2026]. Available from: https://osf.io/fwaud/overview.
Valenzuela, C., Stern, C., & Aromataris, E. (2025). User experience and adoption of automation and AI for evidence synthesis: A scoping review protocol. JBI Evidence Synthesis, 24(3). [CrossRef]
Witkowski, K., Okhai, R., & Neely, S. R. (2024). Public perceptions of artificial intelligence in healthcare: Ethical concerns and opportunities for patient-centered care. BMC Medical Ethics, 25(1). [CrossRef]
Wu, C., Xu, H., Bai, D., Chen, X., Gao, J., & Jiang, X. (2023). Public perceptions on the application of artificial intelligence in healthcare: A qualitative meta-synthesis. BMJ Open, 13(1), e066322. [CrossRef]
Zondag, A. G. M., Rozestraten, R., Grimmelikhuijsen, S. G., Jongsma, K. R., Solinge, W. W. van, Bots, M. L., Vernooij, R. W. M., & Haitjema, S. (2024). The effect of artificial intelligence on patient-physician trust: Cross-Sectional vignette study. Journal of Medical Internet Research, 26(1), e50853. [CrossRef]

Table 1. Inclusion/exclusion criteria.

Domain	Inclusion	Exclusion
Population	Patients or the general public (i.e., lay people). Any age.	Healthcare professionals (HCPs). Medical students, dental students) within the specific context of their education or professional learning/training. Mixed populations (HCPs/students as described above pooled with patients/public and data not presented by patients/public alone).
Concept	Attitudes, perspectives, perceptions, and opinions regarding their own use of a publicly available AI system/application for health information seeking or decision making. Experiences and trust levels associated with using AI for health information seeking. AI is defined as “a set of advanced technologies that enable machines to do highly complex tasks effectively – which would require intelligence if a person were to perform them” (Thomas et al., 2026). Common examples of publicly available AI systems/applications include symptom checkers (e.g., Ada Health), AI-powered health platforms (e.g., Google Health), and generalized platforms (e.g., ChatGPT, Gemini, Claude and AI chatbots).	Studies that assess attitudes, perspectives, perceptions, and opinions of AI developers, clinicians, or system designers for implementation of AI in healthcare. Studies that examine general online health information seeking (e.g., Google, WebMD) without an AI component. Assessing the use of AI systems/applications to deliver interventions for clinical populations.
Context	Seeking health information outside of a clinical setting. Use of AI as a resource for health information seeking or decision making (e.g., symptom checking, understanding a diagnosis, comparing treatment options, preparing for an appointment, evaluating online health claims, finding support with their acute or chronic condition, deciding whether to go to the doctor, figuring out what type of care to seek, preparing for emergency care, etc.).	Testing specific AI systems/applications for accuracy of information. Professional learning or educational contexts. AI used for shared decision making or other purposes during a healthcare encounter. Delivering an intervention to manage a condition (e.g., medication timing for diabetes patients). Use of an AI system/application that only describes the technical performance with a focus on algorithm accuracy, model training, or system architecture which do not include experiences, perceptions, or trust by end-users (patients/public).
Type of Study	Primary research of any design (e.g., quantitative, qualitative, multi-method, mixed-methods).	Reviews (systematic, scoping, etc.). Editorials, commentaries, and opinion pieces.
Publication Status	Studies published in peer-reviewed journals. Grey literature, e.g., theses and dissertations, reports. Conference abstracts (will be used to identify if a complete study has been published).	Unpublished studies (pre-prints). Studies that are not reported in full (i.e., methods and results).
Language/ Country	No restrictions.	N/A
Publication Date	No restrictions.	N/A

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

© 2026 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Copyright: This open access article is published under a Creative Commons CC BY 4.0 license, which permit the free download, distribution, and reuse, provided that the author and preprint are cited in any reuse.