JMIR Publications

ABSTRACT

Background:

Evidence that Artificial Intelligence (AI) may improve melanoma detection has led to calls for increased human-AI collaboration in clinical workflows. However, “AI-based support” may entail a wide range of precise functions for AI. In order to appropriately integrate AI into decision-making processes, it is therefore crucial to understand the precise role that clinicians see AI playing within their clinical deliberations.

Objective:

This study sought to provide an in-depth understanding of how a range of clinicians involved in melanoma screening and diagnosis conceptualise the role of AI within their decision-making, and what these conceptualisations mean for good decision-making.

Methods:

This qualitative exploration used in-depth individual interviews with 30 clinicians predominantly from Australia/New Zealand (26/30) who engage in melanoma detection (17 dermatologists, 6 General Practitioners with an interest in skin cancer, and 7 melanographers). The vast majority of the sample (83%) had interacted with or used, 2D/3D skin imaging technologies with AI tools for screening or diagnosis of melanoma, either as part of testing, through clinical AI reader studies, or within their clinical work.

Results:

We constructed five themes to describe how participants conceptualised the role of AI within decision-making when it comes to melanoma detection. Theme 1 (integrative theme): the importance of good clinical judgement; Theme 2: AI as just one tool among many; Theme 3: AI as an adjunct after a clinician’s decision; Theme 4: AI as a second opinion on unresolved decisions; Theme 5: AI as an expert guide before decision-making. Participants articulated a major conundrum - AI may benefit inexperienced clinicians when conceptualised as an “expert guide”, but over-reliance, de-skilling, and a failure to recognise AI errors, may mean only experienced clinicians should use AI “as a tool”. However, experienced clinicians typically relied on their own clinical judgement, and some could be wary of allowing AI to “influence” their deliberations. The benefit of AI was often then to reassure decisions once they had been reached, by conceptualising AI as a kind of “checker”, “validator”, or in a small number of equivocal cases as a genuine “second opinion”. This perhaps raises queries about the extent to which experienced clinicians then truly seek to “collaborate” with AI or use it to inform decisions.

Conclusions:

Clinicians conceptualised AI-support in an array of potentially disparate ways that potentially impact AI’s incorporation, whilst prioritising the conservation of good clinical acumen. This must be a priority when developing and adopting AI into decision-making process, and our study implores a more focused engagement with users about the precise way, and in what position, they envisage AI being incorporated into their decision-making process for melanoma detection.