Detecting asthma exacerbations in a pediatric emergency department using a Bayesian network

Area of Science:

• Pediatric Emergency Medicine
• Clinical Decision Support Systems
• Health Informatics

Background:

• Asthma care guidelines improve patient outcomes.
• Timely identification of eligible patients is crucial for guideline implementation.
• Electronic health records offer potential for automated eligibility screening.

Purpose of the Study:

• To develop and validate a Bayesian network model.
• To identify pediatric patients eligible for asthma-care guidelines.
• To utilize only data available at the time of patient triage.

Main Methods:

• A Bayesian network was developed and evaluated using clinical data from 3,023 pediatric emergency department visits.
• An independent chart review established a reference standard for guideline eligibility.
• Performance was assessed using area under the receiver operating characteristic curve, sensitivity, specificity, and predictive values.

Main Results:

• The Bayesian network achieved an area under the receiver operating characteristic curve of 0.959.
• At 90% sensitivity, the network demonstrated 88.3% specificity and a 98.8% negative predictive value.
• Positive and negative likelihood ratios were 7.69 and 0.11, respectively.

Conclusions:

• The developed Bayesian network accurately identifies patients eligible for asthma guidelines.
• This automated approach can facilitate the timely initiation of guideline-based asthma care.
• Bayesian networks show promise for improving clinical guideline adherence in pediatric emergency settings.