AI in Skin Cancer Diagnosis: A Tool, Not a Replacement
Getting expert input early is one of the best ways to improve outcomes and reduce costs.
By Dr. Toby Nelson, Community Dermatologist and Co-Founder, Map My Mole
In healthcare, timing is everything - and getting expert input early is one of the best ways to improve outcomes and reduce costs.
Think about how an experienced A&E consultant at the front door can rapidly assess and triage incoming patients. These senior decision-makers can immediately:
This front-loaded expertise prevents delays, avoids unnecessary investigations, and ensures better outcomes.
In contrast, the traditional NHS skin lesion pathway delays access to this level of expertise. Skin conditions account for up to 25% of GP workload1, yet many cases are first seen by a fragmented, overstretched non-specialist workforce - nurse practitioners, newer GPs etc.
Understandably, caution often prevails. The result? Patients may cycle through multiple appointments over several weeks before seeing someone experienced enough to make a definitive call.
The Map My Mole Model: Specialist Input First
At Map My Mole, we’ve designed a smarter pathway that puts expert judgement right at the start:
This is the model that delivers both clinical and system-level impact:
In our own pilot data, consultant-led triage enabled 64% of cases to be managed without secondary care referral — freeing up capacity for those who need it most. (Map My Mole, 2023)
Early-stage melanoma treatment in the UK costs around £9,500 per case, primarily involving simple surgical excision. But if diagnosis is delayed until stage IV, cumulative treatment costs can exceed £214,000 — more than 20 times higher. (British Journal of Dermatology, 2025)
GIRFT data confirms that front-door consultant involvement streamlines patient flow and reduces duplication, improving outcomes while easing burden on overstretched services. (NHS GIRFT Dermatology Report, 2021)
Where Does AI Fit In?
We’re at an inflection point in healthcare.
AI is accelerating - fast - and its role in diagnosis is expanding. But before we hand over too much, we need to ask:
How can AI improve outcomes when it still lacks the ability to consider the whole patient, not just the image?
AI is a powerful assistant, but not yet a replacement for human judgement. It can help with:
✅ Flagging poor image quality
✅ Organising large image sets
✅ Speeding up review workflows
But it lacks clinical nuance.
For example: An image of a suspected carcinoma on the back of a frail care-home resident might trigger an “urgent” flag from an algorithm.
But a consultant might assess it in context (weigh quality of life, co-morbidities, and family wishes) and recommend no intervention — saving unnecessary procedures, stress, and cost. AI may simply label it “suspicious.”
This is what Atul Gawande so powerfully explored in Being Mortal - the tension between medical possibility and what really matters to the person in front of you.
In the age of AI, that tension is even more accentuated.
If we treat AI as the expert, we risk false reassurance, over-referral, and missed context. And that clinical nuance - the ability to hold two truths in mind - is what drives good decision-making.
More importantly, AI alone doesn’t shift the pathway.
Most AI tools are layered into the same point where GPs would already be referring. They don’t bring the expert in earlier - and therefore miss the opportunity to reduce the burden upstream.
The Future is Hybrid
The future lies in combining:
This is how we deliver better outcomes and a more cost-effective system.
This is how we build a smarter, safer dermatology pathway.