Prospective evaluation of a Bayesian Network for detecting asthma exacerbations in a Pediatric Emergency Department

Early detection of asthma exacerbations may allow for automated guideline enrollment. We developed and prospectively evaluated a real-time Bayesian network to predict the presence of acute asthma after patient triage using only routinely available electronic data. 2,006 patients were enrolled, including 153 guideline-eligible patients. The area under the ROC curve was 0.971 (95% CI: 0.955 to 0.981) for identifying asthma patients. The system can be applied to identify guideline eligible patients and remind clinicians to enroll patients into guidelines.