Causal Inference with Artificial Intelligence

Abstract
Cardiovascular infarction, e.g., myocardial infarction (MI), and cerebrovascular infarction, e.g., ischemic stroke (stroke), are leading causes of death and disability in Canada. Both are caused by tissue damage after blocked blood supply, to either the heart or the brain. Self-rated health (SRH) is an individual’s perception of their own health status​, which can be predictive of quality of life, disease re/occurrence, and mortality​. Stroke/MI could impact SRH through a variety of pathways, and psychosocial factors can buffer the challenge of serious health events and lead to vary levels of SRH. In this project, we investigate the causal impact of stroke/MI on SRH, which could inform interventions to maintain good subjective health despite a serious medical event in the context of an aging population​. Utilizing propensity weighting and matching with parametric and machine learning methods, we show that stroke/MI had a negative impact on one’s subjective rating of having very good/excellent health.