Causal analysis for multivariate integrated clinical and environmental exposures data.

Area of Science:

• Computational epidemiology
• Health informatics
• Causal inference in medicine

Background:

• Electronic health records (EHRs) offer valuable real-world patient data for discovering causal links.
• Understanding asthma attack predictors can improve patient outcomes and healthcare strategies.

Purpose of the Study:

• To infer causal relationships from a large-scale EHR dataset for patients with asthma.
• To identify significant predictors of asthma attacks using causal inference methods.
• To simulate interventions on the causal network to assess effects on asthma attack probability.

Main Methods:

• Utilized a large-scale EHR dataset (N = 14,937) including demographics, clinical measures, and environmental exposures.
• Employed causal inference techniques to estimate relationships within the integrated dataset.
• Performed simulated interventions on the inferred causal network to evaluate treatment effects.

Main Results:

• Identified significant predictors of asthma attacks from the EHR data.
• Quantified the causal effects of various factors on asthma attack likelihood through simulated interventions.
• Demonstrated the utility of causal inference on EHR data for understanding disease dynamics.

Conclusions:

• Causal inference on integrated EHR data can effectively identify asthma attack predictors.
• Simulated interventions provide insights into potential strategies for mitigating asthma exacerbations.
• This approach can enhance medical decision-making and generate novel healthcare research hypotheses.