Rule-based machine learning methods for functional prediction
Journal of Artificial Intelligence Research
The problem of bias in training data in regression problems in medical decision support
Artificial Intelligence in Medicine
| Hi-index | 0.00 |
Based on the National Vital Statistics Reports 2001, nearly 115,200 lives are cut short annually because of traumatic injuries and many patients who survive traumatic events have to face the future with life-long disabilities that negatively affect them and their families [29]. Moreover, U.S. trauma center reports suggest that the annual death rate, as well as permanent disabilities, can be reduced when medical treatments are performed rapidly. To provide a certain kinds of medical treatment, a decision has to be made based on the outcomes (discharge status and intensive care unit days). Moreover, predicting the patient's outcome is useful when determining the most reliable transport method: ambulance or helicopter. In this paper, we focus on creating an appropriate decision making process on helicopter transport. Although helicopter transport provides a high level of medical care as well as rapid transport service to trauma patients, the cost of helicopter transport is highly expensive. So there is a question still remains that the use of helicopter transport for trauma patients makes any change in patients' survival and the transport cost. To answer this question, we designed a computer aided decision making system for trauma patients' helicopter transport. The decision making is accomplished by using Decision Tree (DT). Especially, Nonlinear Regression and Classification Tree (CART) is used to create reliable rules. The rules is designed by using a large number of features captured from all available patients' informative attributes. The quality of the extracted rules is also evaluated. Therefore, the computer aided system support making predictions/and recommendations on the outcomes of trauma patients.