Antibiotics are often given to hospital patients where there is initially a concern about an infection, but where later it was found that an infection was not present and antibiotics were not needed after all. Antibiotics are then stopped, but where possible it would be better not to start them in the first place.
Researchers from Sweden have developed a machine learning model capable of predicting which Emergency Department patients are at a low risk of having an infection diagnosed as the cause for their hospital admission. This includes patients with negative tests for infection and excludes the most unwell patients including those needing intensive care to reduce the chances of including patients where an infection may have been missed. The inputs for the model are information that is usually available when a patient is first seen in the Emergency Department. We would like to investigate how well this model works for patients from Oxfordshire.
If the model works well in multiple settings, it could help prevent some low risk patients getting unnecessary antibiotics, which in turn would help to prevent antibiotic resistant developing in future. The analysis in this study will be presented alongside results in patients from Sweden.