Integration of a deep learning basal cell carcinoma detection and tumor mapping algorithm into the Mohs micrographic surgery workflow and effects on clinical staffing: a simulated, retrospective study

Abstract

Background: Staffing shortages and inadequate healthcare access have driven the development of artificial intelligence (AI)-enabled tools in medicine. Accuracy of these algorithms has been extensively investigated, but research on downstream effects of AI integration into the clinical workflow is lacking. Objective: We aim to analyze how integration of a basal cell carcinoma detection and tumor mapping algorithm in a Mohs micrographic surgery (MMS) unit may impact waiting times in the surgical pathology laboratory and on the floor. Methods: Time spent on each task and slide, staff, and histotechnician waiting times were analyzed over a 20 day period in a MMS unit. A simulated AI workflow was created and the time differences between the real and simulated workflows were compared. Results: Simulated addition of the algorithm led to improvements of 64% in slide waiting time (1:03:39 per case), 36% in staff waiting time (59:09 per case), and 25% in histotechnician waiting time (25:27 per case). Limitations: A single MMS unit was analyzed and AI integration was performed retrospectively, rather than in real time. Conclusions: AI integration results in significantly reduced slide, staff, and histotechnician waiting time, which enables increased productivity and a streamlined clinical workflow.

A deep learning algorithm to detect cutaneous squamous cell carcinoma on frozen sections in Mohs micrographic surgery: A retrospective assessment

“…Reducing rate limiting steps to intraoperative margin assessment of cSCC tumours can improve the efficiency and completeness of tumour removal, reducing the burden on laboratory staff while reducing tumour recurrence and repeat procedures. [17][18][19] When evaluating this study, it should be acknowledged that all slides were obtained from a single MMS clinic and scanned images, not slides, were used for training, which may limit generalizability and real-world implementation. Application of this algorithm requires complete, high-quality tissue sections devoid of tears, holes and other artefacts which may preclude histological margin assessment.…”