AI-Supported Haematology Referral Optimisation

The model differs from traditional e-consults because it standardises diagnostic preparation before the first specialist appointment rather than relying mainly on asynchronous specialist advice. The intervention used natural language processing to classify referral notes and automatically generate standardised test order sets before the first specialist haematology appointment. The intervention was implemented through the Casiopea electronic health record and the SCAE primary care database. The standardised test order sets were derived from tests commonly ordered at first appointments and refined by an expert panel. The initiative aimed to make the first haematology appointment diagnostically actionable and reduce low-value follow-up visits for test review.