Key points are not available for this paper at this time.
Abstract Over 25% of patients with suspected skin cancer in England waited over 4 weeks from urgent referral to diagnosis in October 2023. Implementation of artificial intelligence (AI) can augment this pathway to improve timely diagnosis. A prospective, postdeployment, multicentre clinical performance review of AIaMD was performed. AIaMD is a UKCA class IIa-approved artificial intelligence as a medical device intended for use in the screening, triage and assessment of skin lesions suspicious for skin cancer. AIaMD was deployed at four sites as part of the National Health Service AI in Health and Care Award. Patients assessed by the most recent version of AIaMD from April 2022 to November 2023 were eligible for inclusion. Outcomes were confirmed from histology reports for cancerous lesions and histology reports or consultant teledermatology assessment for premalignant and benign lesions. AIaMD assessed 3979 lesions with outcomes confirmed, including 38 melanomas, 68 squamous cell carcinomas, 181 basal cell carcinomas, 5 rare skin cancers, and 612 premalignant lesions such as Bowen disease and actinic keratoses. The negative predictive value – how often AIaMD labelled lesions as premalignant or benign and correctly ruled out cancer – was 99.8% (539 of 540). The sensitivities of AIaMD for melanoma, all skin cancer and premalignancy were 97% (37 of 38), 96.6% (282 of 292) and 86.3% (528 of 612), respectively. The specificity of AIaMD for benign lesions was 74.9% (2118 of 2827). For benign lesions confirmed by biopsy only, the specificity was 27.6% (105 of 381). One site reported a reduction in average wait time to first appointment by 11 days, a 10% reduction in biopsies and a 13% reduction in routine follow-up appointments. This postmarket service evaluation reported the clinical outcomes of AIaMD – an AI device used for assessment of skin lesions. The pathway was sensitive, identifying 282 out of 292 skin cancers, and specific, correctly identifying three out of every four benign lesions assessed and confirmed clinically or histologically. Moreover, implementation of AIaMD was linked to benefits to the pathway such as a reduction of biopsies. The integration of AIaMD into skin cancer diagnostic pathways could significantly improve the accuracy of urgent suspected skin cancer referrals, removing unnecessary specialist review of benign lesions. Part of the deployment period was funded as part of the NHS AI in Health and Care Award, and two of the authors are employed by one of the deployment sites. Three of the authors are employed by the AI provider.
Brewer et al. (2024) studied this question.