Cost effectiveness of the Mia artificial intelligence technology for detection of breast tumours as part of the NHS breast screening service in the UK

Abstract

Abstract In the UK, the NHS National Breast Screening programme, which aims to detect breast cancer at earlier stages, has been shown to be cost-effective. 1,2 The reference standard within the service is for mammograms to be independently double-read3. The double reading is performed by consultant radiologists, consultant radiographers, advanced practitioners, and breast clinicians. If the readers disagree, then arbitration by a single or several readers will take place. This process has reduced false positives and recall rates whilst producing a high level of accuracy, 4 but the process is labour-intensive. In the UK, this has put the service under pressure due to a radiology work force crisis. Artificial intelligence (AI) technology has been suggested as a substitute for a human reader as a solution.5 While such technology has shown to be non-inferior in performance as a second reader6, the minimum requirements needed (effectiveness, set-up costs, maintenance, etc) for such technology to be a cost-effective alternative for use in the NHS, have not been evaluated. To assess the later, we developed a simulation model replicating the UK NHS screening services. Our results indicate that if non-inferiority is maintained, the use of an AI technology as a second reader is a viable and potentially cost-effective alternative for use in a service such as the NHS.