Worldwide, more than 200 million adults undergo major noncardiac surgery each year and the number of such patients is increasing. Each year, more than 10 million adults worldwide have a major cardiac complication in the first 30 days after noncardiac surgery. Major perioperative cardiac complications are important because they account for at least one third of perioperative deaths, result in substantial rates of complications, prolong hospitalization, and increase medical costs. The contributing factors of perioperative cardiac complications are complicated and difficult to define. The sample size of previous research was too small to define the risk factors. However, big data analysis using Taiwan National Health Insurance Research Database (NHIRD) could be ideal approach to find the risk factors of perioperative cardiac complications. Previous NHIRD studies were mostly analyzed by statistic methods. However, new data mining techniques such as artificial neural networks (ANNs), genetic algorithm, and sensitivity analysis have been developed for analysis of large database. ANNs may be superior to statistical methods in some aspects, such as the development of non-linear models, tolerance for missing data, and high adaptation etc. Prediction models based on ANNs and data mining techniques may have better predictive performance than those based on logistic regression analysis. This is a two-year plan. In the first year, we analyze Taiwan NHIRD to build risk assessment models for perioperative acute myocardial infarction in patients with diabetes, hypertension, or stroke. In the second year, we use clinical database which include more parameters such as laboratory values, drug prescriptions, and postoperative complications, to develop risk assessment models for perioperative major cardiac complications. Finally, we evaluated the predictive performance of models and assess the practical value in clinical practice
|Effective start/end date||8/1/17 → 7/31/18|
Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.