Poster Presentation
Friday, July 13, 2007
9:30 AM - 10:15 AM
Friday, July 13, 2007
3:15 PM - 4:00 PM
Building a prediction model for surgical patients of pressure ulcer based on Data Mining and clinical application
Wei-Fang Wang, RN, MSN, Nursing Department, National Cheng Kung University Hospital, Tainan, Taiwan, Chung-Feng Yang, BS, Information Center, National Tainan Institue of Nursing, Tainan, Taiwan, and Vincent S. Tseng, PhD, Department of Computer Science and Information Engineering, National Cheng Kung University, Tainan, Taiwan.
| Learning Objective #1: know how to apply data mining method to nursing practice |
| Learning Objective #2: know what is the prediction model of pressure ulcer in surgical patients by data mining |
The operation patients have to fixed position during the whole operation procedure. Because of the medical and personal factors, surgical patients are high risk group of pressure ulcer. The pressure ulcer incidence rose recently (0.61%), the third season of occurred pressure ulcer of patients was 2.8 times of the first season. Though the operation room carried out multiple measures of improving, incidence went rising instead of falling. The figures showed that reasonable operation time also increased 1.77% relatively. Prevention pressure ulcer equipments have been purchased but still not enough to go around. Patients had to buy artificial skin and other protection materials, because the health insurance does not offer those in Taiwan. It need a scientific data to be a consult indicator for solving the problems. This research intended to build a prediction model based on Data Mining method. Predicting result could keep medical resource make the most effective distribution. It provided patients and families with appropriate suggestion and reduce the pressure ulcer incidence rate of operation patients.
Data collection was used by Operation Patient Pressure Ulcer Assessment Scale. The scale included the demographic data, and perioperative nursing assessment. Up to now there were 1359 event patient data. The performance of data mining was the decision tree (J48). The initial result found leaves number: 9, tree size: 13. The pressure sore risk factors were: 1. patients without artificial skin material and operation time over 6 hours, 2.patients with artificial skin material and poor circulation condition (edema). The correctly Classified Instances was 64.9 %. The study is on going now. Final outcome will apply to design prevention surgical patient pressure ulcer protocol, in order to improve perioperative care quality.