| Summary: | 碩士 === 國立成功大學 === 資訊工程學系碩博士班 === 95 === In this paper, we propose a novel hybrid approach for analysis of patient behaviors by mining the streamed movement log obtained in RFID environments. The proposed approach can discover the regular behavior models of patients using data mining techniques. Moreover, the variation between two behavior models under different periods can also be obtained. We propose three kinds of novel variation evaluation strategies, namely association pattern, sequential pattern, and location change frequency/velocity that utilize the evaluation result to objectively evaluate the variation between two periods on patients. To our best knowledge, this is the first work on evaluating the behavior variation of patients by mining behavior patterns in RFID-based patient monitoring environments. Through a series of experimental evaluations, the proposed hybrid method is shown to be promisingly effective under different system conditions.
|
|---|
