Quantification of the Causal Effects between Laboratory detection, Echocardiogram and Acute Myocardial Infarction

Acute Myocardial Infarction (AMI) is a serve disease with rapid onset, high mortality and complex symptoms. Therefore, the accurate diagnosis and effective treatment of AMI are important but difficult in clinic. In addition to symptoms and electrocardiogram (ECG) features, laboratory detection and echocardiogram (ECHO) provide significant help for physicians. With the support of big data, quite studies utilized machine learning to investigate the association between laboratory detection, ECHO and the onset of AMI. However, most researches just focused on the correlation in data while ignore the causal logic. Therefore in this paper, based on the electronic healthcare record data, the Structural Causal Model is adopted to discover the causal relationships and construct the causal graph. Furthermore, the effective information is utilized to quantify the causal effects between the features and AMI. The results show that regional wall motion abnormality, fasting blood sugar, ejection fraction, erythrocyte sedimentation rate, white blood cell and creatine have significant causal effects with AMI. The proposed method and analysis results can provide guidance for clinical diagnosis and treatment of AMI.