Artificial Intelligence–Informed Exercise Prescription: Perspectives from People with Long-Term Conditions, Carers and Staff

Introduction

Physical activity is central to healthy ageing and long-term condition (LTC) management, yet older adults with multimorbidity, frailty, and fluctuating symptoms face barriers to safe, individualised exercise support in primary care. Exercise prescription (EP) is often limited by time constraints, variable access to specialist input, and clinician confidence in tailoring recommendations. Artificial intelligence (AI)–informed EP tools offer a route to personalised, scalable physical activity support, but raise questions regarding safety, trust, equity, and integration into routine care, particularly for older adults and carers. Evidence on stakeholder perspectives in real-world primary care settings remains limited. This study explored acceptability, benefits, and implementation challenges of AI-based EP among people with LTCs, informal carers, and healthcare professionals.

Methods

A qualitative study using online semi-structured focus groups was conducted. Participants included people living with LTCs, informal carers, and healthcare professionals in primary care or exercise referral, using purposive sampling to capture diverse experience. Four focus groups were conducted (n=13). Data were analysed using reflexive thematic analysis following Braun and Clarke’s six-phase framework, with independent coding and team discussion to enhance rigour.

Results

Six interconnected themes were identified. Acceptability of AI-informed EP depended on transparency, evidence, and clinical oversight. Participants recognised benefits such as personalised exercise support, reduced administrative burden, and improved continuity of care, but raised concerns about safety, accountability, and AI replacing human judgement. Digital exclusion, health literacy, language needs, and socioeconomic barriers were key challenges for older adults and carers. Successful implementation required integration with existing systems, minimal workflow disruption, and hybrid human–digital models with inclusive design.

Conclusion

AI-informed EP may enhance physical activity support for people with LTCs, including older adults, if implemented with transparency, clinical oversight, and inclusive design. Adoption in primary care will depend on alignment with geriatric care priorities and human-centred, integrated delivery models.