Activity State Estimation for Medication Management Support System

2014 Fiscal Year Final Research Report

• Project/Area Number: 25880010
• Research Category: Grant-in-Aid for Research Activity Start-up
• Allocation Type: Single-year Grants
• Research Field: Intelligent informatics
• Research Institution: Yokohama National University
• Principal Investigator: SUZUKI TAKUO 横浜国立大学, 工学(系)研究科(研究院), 研究教員 (80709303)
• Project Period (FY): 2013-08-30 – 2015-03-31
• Keywords: モバイルヘルス / 健康管理 / 物体検出 / 画像処理 / 状態推定 / 行動認識 / 人間支援システム / 人間計測システム

Outline of Final Research Achievements

In this research, an original algorithm that predicts the future state of recipient’s activity was developed to detect a medication error (e.g. a recipient occasionally overdoses or forgets to take medicine) while going out. The future state was estimated based on data measured by a variety of sensors in a smartphone such as an acceleration sensor, a gyro sensor, an ambient light sensor, a touch panel, and a barometer. The data were converted to feature quantities, which do not correlate with each other, by principal component analysis and were entered into a Bayesian network for activity state estimation. The target value of accuracy of estimation was set more than 70 percent, but it was not achieved. The main reason of the bad result was that eating-related states were so difficult to be predicted only by using the above sensors.

Free Research Field

システム工学