“…Machine learning (ML) is being increasingly applied to improve clinical workflow efficiency and has the potential to enhance the accuracy of triage, optimising service allocation. 7 Within triage, ML has the capability to process high dimensionality structured data and the potential to achieve superior performance compared to rule-based algorithms by abstracting complex non-linear patterns between patients’ clinical presentation and their clinical risk. One study proposed an ophthalmic self-triage model using metadata and smartphone images but was tested only on 103 patients, included only 18 possible differentials, and did not consider the potential increase of non-urgent presentations to emergency departments, aggravating professional burden and increasing healthcare costs.…”