Operationalizing Digital Health Equity in Artificial Intelligence-Enabled Patient Decision Aids for Older Adults: Mixed Methods Study

Clinical Snapshot

Key Findings

• Effect Size: Not applicable — qualitative and framework-development study; no quantitative effect sizes reported

• Primary Outcome: Five actionable design recommendations for equitable AI-PDA development: (1) co-design with end users; (2) embrace relationship-centred design; (3) leverage community resources; (4) promote accessible and equitable AI governance; (5) enhance equitable AI through algorithmic fairness

• Nnt Or Sensitivity: Not applicable — this is a design framework study; equity outcomes are conceptual and require future empirical validation through co-designed prototypes and real-world testing

Clinical Application

The recommendations are conceptually feasible but require significant implementation infrastructure: co-design processes with older adult communities, interdisciplinary development teams, algorithmic fairness auditing capacity, and community health worker integration. Resource intensity is high and may be prohibitive in under-resourced settings. No prototype or pilot data are yet available to assess real-world feasibility. In Australia, the relevance is substantial but contextually nuanced. The RACGP supports shared decision-making and patient-centred care in chronic disease management, and the Australian Digital Health Agency's National Digital Health Strategy 2023–2028 explicitly prioritises equity and inclusion. PBS-listed medications for hypertension and diabetes are widely used, and AI-PDAs could theoretically support medication adherence and treatment preference elicitation. However, the Australian context introduces unique equity considerations absent from the Hong Kong study: Aboriginal and Torres Strait Islander health equity, rural and remote digital access disparities, culturally and linguistically diverse (CALD) communities, and My Health Record integration. TGA oversight of AI-enabled software as a medical device (SaMD) under the Software as a Medical Device framework would apply to any AI-PDA deployed clinically. The algorithmic fairness recommendations are particularly relevant given documented biases in AI systems trained on non-representative datasets. Direct application of the Hong Kong-derived framework requires cultural adaptation and Australian-specific co-design processes. Older adults (65+ years) with chronic conditions, particularly hypertension and type 2 diabetes, who may engage with AI-enabled digital health tools in primary care or chronic disease management settings

Abstract

BACKGROUND: Artificial intelligence-enabled patient decision aids (AI-PDAs) hold promise for supporting older adults with chronic diseases in accessing personalized health information, clarifying preferences, and engaging in shared decision-making. Achieving equity in their design requires attention to the complex health care and digital contexts in which these tools are used. While the Digital Health Equity Framework (DHEF) provides a conceptual foundation, practical strategies for its application remain limited. OBJECTIVE: This study aimed to identify equity-related determinants and generate actionable design strategies for applying the DHEF to AI-PDAs for older adults. METHODS: A mixed methods study was conducted. Semistructured interviews were conducted with older adults living with hypertension and/or diabetes, health care providers, and medical students to explore equity determinants relevant to AI-PDAs. In parallel, a review of reviews synthesized existing evidence on approaches to addressing these determinants. Interview findings and review findings were integrated through an iterative mapping process conducted by the research team and refined through multidisciplinary expert consultation involving medicine, public health, social services, and computer science. RESULTS: A total of 33 stakeholders were interviewed, including 15 older adults, 8 health care providers, and 10 medical students. Thirteen reviews were included in the umbrella review. The integrated synthesis identified equity determinants spanning individual, interpersonal, community, and societal levels across both the health care and digital environments, together with cross-level concerns related to algorithmic fairness. These findings informed 5 recommendations for equitable AI-PDA development: (1) co-design with end users to address their needs, (2) embrace relationship-centered design, (3) leverage community resources to improve support, (4) promote accessible and equitable artificial intelligence (AI) governance in society, and (5) enhance equitable AI through algorithmic fairness. Together, these recommendations provide practical guidance for design, pilot testing, implementation, and evaluation. CONCLUSIONS: By integrating stakeholder perspectives with synthesized review evidence, this study extends the DHEF from a primarily conceptual framework toward a more practice-oriented approach for AI-PDAs for older adults with chronic disease. Health care settings serve as a mediating sociotechnical context where AI tools may either support or constrain equitable care participation. The findings underscore the need for interdisciplinary collaboration to align technological innovation with equity-oriented design. Future work should focus on co-designed prototypes, real-world testing, and measurable equity outcomes.

References

1. 1.Tian, C. Y., Yang, X., Wang, K., Cheung, A. W.-L., Ma, J. C.-H., Lu, C., Yu, J. C.-Y., Chan, C. Y., Chen, J., Ouyang, K., Lin, I. W.-K., Pang, T. H.-C., Zhao, S., Wang, Y., & Wong, E. L.-Y. (2026). Operationalizing digital health equity in artificial intelligence-enabled patient decision aids for older adults: Mixed methods study. Journal of Medical Internet Research. https://doi.org/10.2196/89011