Using classification tree and logistic regression methods to diagnose myocardial infarction.

摘要: Early and accurate diagnosis of myocardial infarction (MI) in patients who present to the Emergency Room (ER) complaining chest pain is an important problem emergency medicine. A number decision aids have been developed assist with this but not achieved general use. Machine learning techniques, including classification tree logistic regression (LR) methods, potential create simple aids. Both a (FT Tree) LR model LR) predict probability that patient having MI based solely upon data available at time presentation ER. Training came from set collected Edinburgh, Scotland. Each was then tested on separate Edinburgh set, as well different hospital Sheffield, England. Previously published models, Goldman tree[1] Kennedy equation[2], were evaluated same test sets. On results showed FT Tree, LR, performed equally well, ROC curve areas 94.04%, 94.28%, 94.30%, respectively, while Tree's performance significantly poorer, area 84.03%. The difference between first three models significant beyond 0.0001 level. Sheffield (p > = 0.17), Tree again outperformed 0.006). Unlike previous work[3], study indicates trees, which certain advantages over may perform MI.