An Improved Medical Decision Model Based on Decision Tree Algorithms
Guiduo Duan, Dan Ding, Yuan Tian, Xiaoming You · 2016
In the medical decision field, the conventional analysis methods of medical insurance data are lack of flexibility and efficiency. When coping with huge and redundant medical dataset, it is difficult to extract the candidate attributes if only using some limited professional knowledge. Therefore, the correlations between the medical insurance cost and relevant factors (e.g. diagnosis results) need to be mined by techniques such as the FP-Growth algorithm, which uses an extended prefix-tree structure for building attributes automatically, storing compressed and crucial information. In this work, we propose an improved decision tree algorithm to effectively model medical data, which includes processes of pre-pruning and post-pruning. Classification based on three criteria such as accuracy, stability and complexity are then utilized to improve the effectiveness of our approach. Simulation results show that our proposed model can improve the decision making with higher flexibility and efficiency.