This is the first of a 3-phase AI implementation, to evaluate the performance for the specific population before further implementation is undertaken and understand whether there are any adjustments required. 
A cohort of adult (over 18 years) patients referred from General Practice and Outpatient for a frontal chest radiograph (CXRs) was collected from a single National Health Service Trust retrospectively from examinations conducted from November 2022 - January 2023. The reference index was established by consensus between at least two consultant radiologists.
The aim of the AI tool is to categorise the radiographs into 1 of 4 categories of findings. Category 1 was any findings related to lung cancer, category 2 was urgent non-cancerous findings, category 3 was any lines and tubes and finally category 4 was chronic or no AI findings. In the analysis, categories 1 & 2 were deemed remarkable and categories 3 & 4 were unremarkable.
The analysis included basic performance metrics: sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV).