Artificial intelligence (AI) is transforming healthcare, yet the populations shaping these innovations remain unevenly represented. A recent systematic review published in JMIR Human Factors (Shiroma & Miller, 2025) highlights a stark gap: rural older adults are largely invisible in AI-driven health research.
The demographic shift toward an ageing population is especially pronounced in rural areas, where older people may experience unique challenges in accessing emergency health care. AI and telehealth technologies are frequently promoted as potential solutions, but without meaningful inclusion of rural populations in training data, these systems risk amplifying inequities rather than reducing them.
Following PRISMA guidelines, the authors systematically searched seven major databases for empirical, English-language studies published between 2013 and 2023. To be included, studies needed to involve adults aged 65 years and older and report on the application of AI in rural health contexts. 23 studies were included.
Representation of rural older adults was limited. Most studies drew on data from large, urban health systems, and only 30% focused exclusively on rural populations. In many cases, rurality was not even reported as a variable.
Machine learning was the most common AI approach, supporting predictive models for depression risk, cognitive decline, “active ageing,” suicidal ideation, stroke, and coronary heart disease. Deep learning methods were applied to image-based screening, such as retinal lesion detection in rural populations. Robotics featured in studies on telerehabilitation for rural veterans and remote consultation through robot-assisted echocardiography. One study described the participatory design of socially assistive robots with older adults.
Thematic analysis revealed three key themes: Numbers over narratives, Efficacy over Impact and Deepening Disparities. Most research privileged numbers over lived experiences, emphasised efficacy over real-world impact, and overlooked rurality as an analytical lens altogether. Nearly half the studies were conducted in China, raising questions about the global relevance of their findings to countries like Australia and Aotearoa New Zealand.
AI tools are only as useful as the data that train them. When rural older adults are excluded, predictive algorithms may fail to recognise the presentations and comorbidities most common in rural emergency care. The authors call for rurality to be treated as a key demographic variable, like age or sex, and for funders and ethics committees to prioritise inclusion of rural participants and transparent geographic reporting.
For RRECN members, the message is clear: by contributing rural data, engaging in research design, and advocating for equity in digital health, we can help ensure that AI in healthcare works for all – not just those who live near a tertiary hospital.
First published in the RRECN Newsletter (November 2025).
Working in rural emergency care? Join the RRECN network