Fall Detection

Aging society has approached in Taiwan, healthcare for the elderly has gradually gained its importance. Early and accurate detection of the older people’s accidental falls can significantly shorten the time to receive instant and proper treatment. This can also lessen the avoidable harm resulting from the delay of sending the patients to the hospital, and thus improve the quality of healthcare. The study proposes a fall detection mechanism which can quickly and accurately detect falls of any kind. By utilizing the built-in 3 axis-accelerometers in Android-based smartphones, a fall detection mechanism was constructed from the triaxial variations and phone’s body change of pre-falls and post-falls. Such mechanism was later realized by an Android application. When the smartphone running the application is placed in 6 positions, and 5 using method of bags, the mechanism can effectively detect real falls, including forward, backward, rightward, and leftward falls. On the other hand, non-fall activities such as running, walking, sitting down, squatting, going upstairs downstairs, and other special movement will not be mistakenly recognized as real falls. The average sensitivity of the presented detection mechanism is 91.5%, and the average specificity is 97.5%. These results prove the effectiveness of the presented mechanism. Aging society has approached in Taiwan, healthcare for the elderly has gradually gained its importance. Early and accurate detection of the older people's accidental falls can significantly shorten the time to receive instant and proper treatment. This can also lessen the avoidable harm resulting from the delay of sending the patients to the hospital, and thus improve the quality of healthcare. The study proposes a fall detection mechanism which can quickly and accurately detect falls of any kind. By utilizing the built-in 3 axis-accelerometers in Android-based smartphones, a fall detection mechanism was constructed from the triaxial variations and phone's body change of pre-falls and post-falls. Such mechanism was later realized by an Android application. When the smartphone running the application is placed in 6 positions, and 5 using method of bags, the mechanism can effectively detect real falls, including forward, backward, rightward, and leftward falls. On the other hand, non-fall activities such as running, walking, sitting down, squatting, going upstairs downstairs, and other special movement will not be mistakenly recognized as real falls. The average sensitivity of the presented detection mechanism is 91.5%, and the average specificity is 97.5%. These results prove the effectiveness of the presented mechanism.