Implementation challenges of artificial intelligence (AI) in primary care: Perspectives of general practitioners in London UK

Abstract

INTRODUCTION: Implementing artificial intelligence (AI) in healthcare, particularly in primary care settings, raises crucial questions about practical challenges and opportunities. This study aimed to explore the perspectives of general practitioners (GPs) on the impact of AI in primary care. METHODS: A convenience sampling method was employed, involving a hybrid workshop with 12 GPs and 4 GP registrars. Verbal consent was obtained, and the workshop was audio recorded. Thematic analysis was conducted on the recorded data and contemporaneous notes to identify key themes. RESULTS: The workshop took place in 2023 and included 16 GPs aged 30 to 72 of diverse backgrounds and expertise. Most (93%) were female, and five (31%) self-identified as ethnic minorities. Thematic analysis identified two key themes related to AI in primary care: the potential benefits (such as help with diagnosis and risk assessment) and the associated concerns and challenges. Sub-themes included anxieties about diagnostic accuracy, AI errors, industry influence, and overcoming integration resistance. GPs also worried about increased workload, particularly extra, unnecessary patient tests, the lack of evidence base for AI programmes or accountability of AI systems and appropriateness of AI algorithms for different population groups. Participants emphasised the importance of transparency, trust-building, and research rigour to evaluate the effectiveness and safety of AI systems in healthcare. CONCLUSION: The findings suggest that GPs recognise the potential of AI in primary care but raise important concerns regarding evidence base, accountability, bias and workload. The participants emphasised the need for rigorous evaluation of AI technologies. Further research and collaboration between healthcare professionals, policymakers, and technology organisations are essential to navigating these challenges and harnessing the full potential of AI.

Ähnliche Arbeiten