Please use this identifier to cite or link to this item:
Title: 基於人體狀態及交通運輸模式識別技術在智慧型手機上發展減重偵測系統
Developing a Weight Loss Monitoring System on Smartphone Using Human Behavior and Transportation Detection Technology for Healthcare
Authors: 陳志銘
Chih-Ming Chen
Chin-Ming Hong
Keywords: 智慧型手機
Weight loss monitoring system
Human activity detection
Transportation detection
Issue Date: 2010
Abstract: 本研究在內建三軸加速度感測器的智慧型手機平台上成功發展具識別人體活動狀態及交通運輸模式的方法。並且基於人體狀態及交通運輸模式識別結果發展一套個人熱量紀錄器,能夠提供使用者24小時的熱量消耗監測記錄並提供簡單的統計圖表,讓使用者知道每天的活動量及其所消耗熱量,作為使用者實施減重健康照護時的有效輔助工具。 智慧型手機平台採用HTC HERO,三軸加速度感測器資料收集則採用Accelogger,並且利用Weka發展人體狀態及交通運輸模式預測模型,在測試比較決策樹、邏輯迴歸、類神經網路、支持向量機以及貝氏分類器等五種分類演算法後,以決策樹得到最佳的分類準確率。因此本研究採用決策樹當成人體狀態及交通運輸模式分類預測工具,分別可識別出走路、慢跑、搭公車、搭捷運以及靜止狀態,且準確率達到97.1954%。發展完成之系統,在隨機抽樣使用者試用後,顯示本研究發展之系統對於應用於個人減重健康照護具有實際應用價值。
The study presents a novel scheme, which can accurately identify human activities such as running, walking, stillness and transportation statuses such as taking bus and MRT, based on samrtphone with build-in tri-axial accelerometer. Moreover, a weight loss monitoring system with precisely calculating consuming calories was successfully developed for healthcare in daily life based on the above-mentioned technologies of identifying human activities and transportation statuses. In the study, the HTC HERO smartphone with build-in tri-axial accelerometer and Android operating system was adopted as platform to develop the proposed weight loss monitoring system. Additionally, an application program named Accelogger with Fast Fourier Transformation (FFT) was employed to sense data of human activities and transportation statuses from tri-axial accelerometer for collecting training data and performing feature selection to model a prediction model. Meanwhile, the study applied Weka, which is a data mining tool, to implement the proposed prediction model of identifying human activities and transportation statuses. After comparing five well-known pattern classification schemes in Weka, decision tree has the best performance in terms of classification accuracy rate, and the classification accuracy rate on predicting three human activities and two transportation statuses is up to 97.1954%. Therefore, the study selects decision tree as prediction model for the proposed weight loss monitoring system. Finally, the proposed weight loss monitoring system was tested by six users who have different life styles during two weeks and an interview was performed to evaluate the satisfactory degree after they used the proposed system for weight loss monitoring. The experimental results show that the proposed weight loss monitoring system is indeed helpful to users to set a weight loss plan based on their self-regulated abilities.
Other Identifiers: GN0697700318
Appears in Collections:學位論文

Files in This Item:
There are no files associated with this item.

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.