Systematic review highlights AI-supported triage, nursing education, and workforce strategies for underserved communities
In a recent interview with
The review followed PRISMA guidelines and evaluated 32 studies published between 2019 and 2024 examining AI-assisted dermatology, teledermatology, and workforce strategies in provider shortage settings. Across the literature, AI-supported technologies consistently reduced wait times, streamlined referrals, improved diagnostic performance, and supported safe task shifting to primary care clinicians, NPs, PAs, and nurses.4
For Madison, one of the review's biggest messages extends beyond AI itself.
The project originated through
"It's not that I expect every nurse or NP to be an expert when it comes to AI... but I do want them to feel confident to sit at any table and have the conversation," Madison said.
There's thousands of drugs. They don't know every single drug, but they understand drug classes... and that's the same mission when it comes to technology and AI specifically. — Kimberly Madison, DNP
Rather than encouraging clinicians to master every AI platform, Madison believes
The review found AI-assisted technologies consistently improved dermatology triage while reducing unnecessary referrals.
Across included studies, AI-supported diagnostic tools demonstrated sensitivities ranging from approximately 85% to 97% for suspicious skin lesions. Several studies reported dermatologist-level diagnostic performance when AI was used alongside clinician judgment, while one study demonstrated a 53% reduction in unnecessary in-person dermatology referrals through AI-assisted image analysis.13
For APPs practicing in primary care or underserved communities, Madison and Trevino see AI as a clinical decision-support tool, not a replacement for clinician expertise.1
The review highlights opportunities for AI-generated differential diagnoses, confidence scores, and visual explanations to help clinicians narrow diagnoses, improve referral decisions, and strengthen diagnostic confidence, particularly among providers with limited dermatology training.
Although access disparities are often associated with rural medicine, Madison said geography tells only part of the story.
"When you first hear access, you think rural communities... but also in a densely populated area like Houston, you can still struggle with access," — Kimberly Madison, DNP
She noted that limited access may also reflect differences in clinician expertise,
Trevino said the review reinforced what she has observed throughout more than a decade in dermatology practice.1
"What stood out most was... how accurate the AI was in diagnosing and identifying suspicious lesions such as skin cancer, It really showed me that this is a problem that is broader than I initially expected." — Jade Trevino, BSN, RN,
She recalled working in a pediatric dermatology clinic where wait times approached one year, particularly for patients insured through Medicaid or without insurance.
Although the findings support broader adoption of AI-assisted dermatology, Madison and Trevino caution that implementation challenges remain.
The review identifies image quality, algorithmic bias, limited diversity in AI training datasets, and the need for more real-world implementation research as ongoing barriers to wider adoption.
Rather than viewing AI as a replacement for clinicians, the authors argue technology should enhance clinical judgment while allowing dermatologists, NPs, PAs, and nurses to practice more efficiently and improve access for patients who need specialty care most.
REFERENCES:
