A study of A Sensing of The Level of Daily Activities : LDA Using Neural Networks for Rehabilitation and Care
Project/Area Number |
16500369
|
Research Category |
Grant-in-Aid for Scientific Research (C)
|
Allocation Type | Single-year Grants |
Section | 一般 |
Research Field |
Rehabilitation science/Welfare engineering
|
Research Institution | Kumamoto National College of Technology |
Principal Investigator |
NAGATA VV Masanobu Kumamoto National College of Technology, Department of Electronic Control, Associate Professor, 電子制御工学科, 助教授 (40370051)
|
Project Period (FY) |
2004 – 2005
|
Project Status |
Completed (Fiscal Year 2005)
|
Budget Amount *help |
¥3,600,000 (Direct Cost: ¥3,600,000)
Fiscal Year 2005: ¥1,000,000 (Direct Cost: ¥1,000,000)
Fiscal Year 2004: ¥2,600,000 (Direct Cost: ¥2,600,000)
|
Keywords | Rehabilitation / Care / Daily Activity / Neural Network / Acceleration Sensor / 傾斜センサー |
Research Abstract |
In the field of rehabilitation or care, it is extremely important that medical staff clearly understand the appearance of daily activities (from now it will be referred to as "the Level of Daily Activities : LDA," and a definition of LDA is given in next chapter) of each subject, especially the five appearances of lying, sitting, standing, walking and wheelchair-driving (lying, sitting and standing will be referred to as "Posture in the Level of Daily Activities : PLDA, and walking and wheelchair-driving will be referred to as "Action in the Level of Daily Activities : ALDA") in order to decide the method of medical treatment, the rehabilitation program or care plan. Some judgment method of LDA with if-then rules has proposed. In those methods, general-use threshold values of sensor data are found respectively, and then LDA is judged by comparing those data to those thresholds with if-then rules. These methods can be used for many subjects. However, it is known that the symptoms of subjects are usually different from each other depending on the field of the rehabilitation or care. If the posture or action in LDA to a subject is over the range of the general-use threshold values, the judgment of LDA with the Evaluating Based on If-then Rules might be difficult. In this paper, a new evaluating method of LDA is proposed with neural networks in order to solve the problem of this individual variation. In the new evaluating method, a neural network acquires specific rules for a specific subject by learning with the measurement signals of the subject, and LDA of the subject is judged with the neural network which has obtained the specific rules. Thus each neural network is used for each corresponding subject. We call this method "Evaluating Based on Neural Networks". The effectiveness of the proposed Evaluating method is shown by judging it with the actual measurement signals of actual subjects.
|
Report
(3 results)
Research Products
(7 results)