IEEE Consumer Electronics Magazine - January/February 2020 - 56

Smart Home Healthcare

Figure 1. Hardware system configuration of
vibration feedback control and muscle activity
monitoring.

accelerometers were applied to both of each
subject's distal tibia. The accelerometers were
sampled at 148 Hz and integrated with a wireless
connection to the Delsys EMG Base Station. The
accelerometers were used to define the walking
gait cycle.10 The hardware system configuration
and sensor attachment on subjects were shown
in Figures 1 and 2.

Maximum Voluntary Isometric Contraction
(MVIC)
Four EMG electrodes were attached (one
each over the VL and VMO on the left and right
legs) and secured with a medical-grade tape.
Each subject was then directed to sit in a
chair, which was height-adjusted to ensure that
the calves dangled naturally. At the start of the
test, the subjects were asked to try to straighten
their calves while another person held their
ankles and applied force in the opposing direction. The test lasted 3 seconds, during which the
strongest EMG signal received was considered
the maximum voluntary isometric contraction of
the muscles.
Data Processing
Raw EMG signals were collected at 2,052 Hz
by the Trigno Wireless EMG unit. Data were sent
in real time to a computer via USB port. We
applied a 10-500-Hz 4th-order Butterworth bandpass filter to the EMG data; full-wave rectification
and root mean square with a moving window of

Figure 2. EMG (left) and Accelerometer (right)
sensor attachment on the subject.

5% of the sampling rate were also applied to the
signal.
MVIC was used for EMG data normalization.
The MVIC data were collected separately before
the walking trial to avoid fatigue. The corresponding algorithms were implemented in MATLAB R2017b (MathWorks Inc., Natick, MA, USA).
The processed signal was then divided into
separate strides; stride times were determined
by additional accelerometers placed on each
side of the distal tibia.
Stride segmentation was calculated from vertical accelerometer data on the distal tibia sensor. Vertical acceleration data from each side
were filtered using a low-pass 4th-order Butterworth filter with a cutoff frequency of 5 Hz.
Then, peak detection was used to detect the
local maxima of the filtered acceleration data.
Consecutive local maxima constituted one walking stride cycle.

Baseline and Abnormal Signal Judgment
We used the VL/VMO ratio to determine
whether a subject's muscle status was normal.
All the received EMG signals were divided by
each subject's MVIC to determine the VL/VMO
ratio during muscle activation. If the activation
threshold is set too low, some physiological signals can cause interference; conversely, if the
threshold is too high, then some of the EMG signals generated by the movement can be dismissed.11 In this study, we set 5% (signal/MVIC)
and 30% as limits to avoid noise interference.

IEEE Consumer Electronics Magazine

IEEE Consumer Electronics Magazine - January/February 2020

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - January/February 2020