Gait phase detection in able-bodied subjects and dementia patients.

Accurate detection of gait phases allows identification of specific functional deficits at each phase of the gait cycle for motor function assessment. This paper proposes a robust gait phase detection method to identify the seven gait phases in overground walking for normal and pathologic gaits. Four inertial sensors are used to obtain knee angles, tibia angles and feet angular rate patterns in the sagittal plane. The key events segmenting the gait cycles are searched using an adaptive threshold in adaptive searching intervals to make sure it works well for different subjects with high variation in cadence and step length during walking. The subjects involved in this study are categorized into three groups: five healthy adult subjects, two healthy elderly subjects and two severe dementia patients. The experimental results have shown our method can reliably detect all gait phases for able-bodied subjects and dementia patients without subject-specific calibration.