Isabel - A Web-based differential diagnostic tool for emergency paediatrics: results of an initial evaluation
P Ramnarayan, A Tomlinson, A Rao, J Britto

Objectives: To test the accuracy and the safety of a Web-based differential diagnostic tool (Isabel) for a set of case histories collected during two phases of evaluation, Phase I: hypothetical cases provided by clinicans and Phase II: real clinical presentations to emergency paediatric departments.

Design: Prospective, descriptive study.

Setting: Acute paediatric units in 2 teaching and 2 district general hospitals in the South-East of England.

Materials: Sets of summary clinical features and their expected diagnoses from Phase I (hypothetical cases provided by participating clinicians in August 2000) and final diagnoses from Phase II (real children presenting to the participating acute paediatric units between October and December 2000).

Main outcome measures: Presence in the Isabel output list of a) the expected or final diagnosis and b) all clinically significant diagnoses for each case, as decided by an expert panel.

Results: 99 hypothetical cases from Phase I, and 100 real-life case scenarios from Phase II were included in the study. Isabel displayed the diagnosis expected by the clinician in 90/99 hypothetical cases (91%). Cases from Phase II evaluation covered a range of paediatric specialities and represented 55 unique final diagnoses. Isabel displayed the final diagnosis in 83/100 real cases (83 %) and the complete set of all clinically significant diagnoses for each case in 73/100 (73%) cases.

Conclusion: Isabel was safe and accurate in generating differential diagnoses for a variety of real as well as hypothetical case scenarios.