Artificial intelligence (AI) allows computers to mimic human cognitive functions, such as deep learning, problem solving and creativity. In recent years, artificial intelligence has stimulated a number of innovations in the medical field. Clinical applications of AI are most advanced in image- and signal-intensive disciplines, including radiology, dermatology, and critical care. It is in these contexts that the performance of AI algorithms for many tasks now meets or exceeds that of individual clinicians.
Primary care supports the health of all members of society and is poised to realize the benefits of AI at scale. Primary care electronic health records contain longitudinal data spanning illnesses, care settings, socioeconomic circumstances, and life experiences. Applications of artificial intelligence to these and other related data (for example, from wearable devices) can enable proactive care and clinical decision support in primary care.
Examples of AI applications targeting primary care under development in the public sector include automating real-time clinical decision checks against chronic disease guidelines, detecting signs of dementia, and predicting outcomes such as non-elective hospitalizations. Despite optimism for the use of AI in primary care, no comprehensive review of the contribution AI has made has been undertaken so far, and there is little guidance on how it should proceed.
A qualitative study brought together a diverse group of primary care physicians, patients and other healthcare professionals, including healthcare system leaders, to prioritize AI when applied in the primary care setting. The study used deliberative dialogue, a participatory method initially developed to improve deliberative democracy by bringing together people affected by an issue to advise decision makers. The method has been adapted to set agendas in various contexts, including health system planning.
The authors identified the following three themes: (1) priority applications of AI in primary care, (2) the impact of AI on the roles of primary care providers, and (3) considerations for training health professionals in AI.
Shared values included health equity, patient-centred care, patient safety, affordability, and continuity of care. Patients and providers have identified strikingly similar priority applications for AI and similar concerns about the impact of AI on care.
Patients and providers have agreed that the highest priority AI applications in primary care involve support for clinical documentation, practice operations and triage, as well as support for clinical decision-making.
However, patients and healthcare professionals have perceived applications in these areas as an immediate and high risk to patient safety, due to unresolved issues of algorithmic bias and the current paucity of evidence regarding the safety and efficacy of the ‘AI.
A sense of fear prevailed about triage tools, which, if designed and implemented without input from key stakeholders, could disrupt continuity or limit access to those unable to use the technology.
Influence vendor roles
When it comes to the impact of AI on the roles of healthcare professionals, most patients and healthcare professionals were not convinced that AI will ever completely replace healthcare professionals, especially in the context of clinical decision-making. This skepticism arose from the idea that the patient-provider relationship is intrinsically human and is at once the defining feature and enabling mechanism of patient-centred primary care.
Several participants were particularly pessimistic that AI tools could fairly or comprehensively consider the social and economic factors influencing care.
With regards to artificial intelligence and its impact on physician training and skills education, there was widespread concern that future generations of providers could suffer from de-skills (i.e., de-skills over time resulting from automation) if the design of artificial intelligence applications or AI training for healthcare professionals does not preserve core competencies that promote patient safety or patient centricity.
Vendors and system leaders have identified the following three priority areas for professional training and continuing education: basic AI literacy, critical evaluation of algorithms, and workflow integration.
The overall findings offer an agenda for applying AI in primary care that builds on the shared values of patients and providers. The authors of the research propose a new paradigm in which, from the concept stage, AI developers work with interdisciplinary teams involving primary care end users as design partners to develop AI-driven tools that address the needs more urgent than patients and providers, unmet needs.
This article was translated by Univadis Italia, which is part of the Medscape Professional Network.