A decision tree-based approach for cardiovascular dysautonomias diagnosis

Terms as knowledge Discovery from Databases (KDD), Data Mining (DM) and Machine Learning (ML), gain from day to day, an increasing significance in medical data analysis. They permit the identification, evaluation, and quantification of some less visible, intuitively unpredictable, by using generally large sets of data. Researchers have long been concerned with applying DM tools to improve data analysis on large data sets. DM has been increasingly used in medicine, particularly in cardiology. In fact, data mining applications can greatly benefits all parts involved in cardiology. Autonomic nervous system (ANS) is the part of the nervous system that is involved in homeostasis of the whole body functions. A malfunction in this system can lead to a cardiovascular dysautonomias. Thereby, a set of dynamic tests are adopted in ANS units to diagnose and treat patients with cardiovascular dysautonomias. In this paper, a case study was performed in order to construct a cardiovascular dysautonomias prediction system using data mining techniques and a dataset collected from an ANS unit of the Moroccan university hospital Avicenne. The prediction system is a decision tree-based classifier that was developed using C4.5 decision tree algorithm to automate the analysis procedure of ANS's test results and make it easier for specialists. The performance of the generated decision trees was evaluated and the results obtained achieved high accuracy rates which were very promising. In addition, a clinical validation of the developed system was carried out on new patients. In fact, a prototype of the developed system was implemented on JEE platform and deployed in the ANS unit so as to be validated clinically. The results were analyzed and thus the prototype was approved to be highly accurate, interpretable, time saving and easy to use.