Key points are not available for this paper at this time.
Abstract Dermatology services are facing rising demand with limited workforce and resources. Artificial intelligence as a medical device (AIaMD) may offer a way to expand capacity and improve patient outcomes. The aim of this project was to develop a cost–utility health economic model of an AIaMD used for the screening, triage and assessment of lesions suspicious for skin cancer using dermoscopic images. This report focuses on the triage of patients referred by general practitioners on the dermatology urgent suspected cancer pathway. The AIaMD assessment takes place in a photo clinic and the images and medical history are subsequently assessed by a specialist (teledermatology). The aim of the AIaMD triage is to identify patients with noncancerous lesions who could be discharged. Four diagnostic strategies are compared: face-to-face assessment (comparator), teledermatology (comparator), AIaMD, and AIaMD with a dermatologist second read (AIaMDSR). A decision tree with Markov models at the terminal nodes was constructed to estimate the long-term costs and outcomes in quality adjusted life-years (QALYs) of each intervention. Three diagnostic categories were considered: (i) high-risk cancer (including melanoma, squamous cell carcinoma and rare cancers), (ii) basal cell carcinoma and (iii) noncancerous (benign or precancerous) lesions. The model has a life-time horizon (up to age 100 years), it takes National Health Service and personal social services perspectives, and the costs and benefits are discounted at an annual rate of 3. 5%. One-way and probabilistic sensitivity analyses were conducted to quantify any uncertainty in the model. Modelling shows that the AIaMDSR and AIaMD interventions are more effective and less costly than face-to-face assessment and teledermatology. The cost per QALY threshold set by the National Institute for Health and Care Excellence is £20 000–£30 000. The most cost-effective interventions were DERMSR at a willingness-to-pay (WTP) threshold of £30 000 and DERM at WTP threshold of £20 000. The results were confirmed by the sensitivity analysis. Cost savings are due to reduced face-to-face appointment and biopsy costs. Savings are calculated as £52 per patient, £259 000 for an average trust or £35 million for NHS England. QALY benefits are due primarily to reduced patient anxiety and biopsies. The relatively small changes to false negatives between the interventions and standards of care do not have a significant an impact. Capacity modelling showed that only the AIaMD intervention is likely to significantly reduce total dermatologist workload compared with face-to-face care, with a reduction of 15%. Some parameter uncertainty exists, reflecting a paucity of baseline data, which will be investigated further with refinement of the model as more data becomes available. This modelling work was funded as part of the NHS AI in Health and Care Award. Three of the five authors are employed by the AI provider.
Zhelev et al. (2024) studied this question.