Performance of a web-based clinical diagnosis support system for internists

Area of Science:

• Medical Informatics
• Clinical Decision Support Systems
• Diagnostic Accuracy

Background:

• Clinical decision support systems (CDSS) aim to enhance medical diagnosis and minimize errors.
• Previous CDSS often presented usability challenges and had limited efficacy in complex cases.

Purpose of the Study:

• To evaluate the diagnostic sensitivity and speed of the Isabel web-based CDSS.
• To assess its performance in complex adult medical cases.

Main Methods:

• Tested 50 Internal Medicine cases from the New England Journal of Medicine.
• Input methods included manual key findings entry (3-6) or pasting the entire case history.
• Evaluated the frequency of correct diagnosis suggestion and data entry/retrieval speed.

Main Results:

• Isabel suggested the correct diagnosis in 96% of cases with key findings entry (48/50).
• It suggested the correct diagnosis in 74% of cases when the entire history was pasted (37/50).
• Results were delivered within 2-3 seconds, with manual entry under a minute.

Conclusions:

• The Isabel CDSS demonstrated high diagnostic accuracy, especially with key findings.
• The system is rapid and effective for complex cases in this study.
• Further evaluation in clinical practice is warranted.