Patient-Led Skin Cancer Teledermatology without Dermoscopy during the Covid pandemic: Important lessons for the development of future patient-facing teledermatology & AI-assisted self-diagnosis.

Abstract

MySkinSelfie was a mobile phone application for skin self-monitoring enabling secure sharing of patient-captured images with healthcare providers. This retrospective study assessed MySkinSelfie's role in remote skin cancer assessment at two centres for urgent (melanoma & squamous cell carcinoma) and non-urgent skin cancer referrals, investigating the feasibility of using patient-taken images without dermoscopy for remote diagnosis. Total number of lesions utilising MySkinSelfie was 814 with mean age of 63. Remote consultations reduced face-to-face appointments by 90% for basal cell carcinoma and 63% for two-week-wait referrals. Diagnostic concordance (consultant vs histological diagnosis) rates of 72% and 83% were observed for basal cell carcinoma (n=107) and urgent skin cancers (n=704), respectively. Challenges included image quality, workflow integration and lack of dermoscopy. Higher sensitivities have been observed in recent Artificial Intelligence (AI) algorithms employing dermoscopy. While patient-taken images proved useful during the pandemic, further research is needed to explore the feasibility of widespread patient-led dermoscopy to enable direct patient-to-AI diagnostic assessment.