Skip to main content

Table 4 Detailed descriptions of newly designed features

From: Machine learning classification of multiple sclerosis patients based on raw data from an instrumented walkway

New feature design

Toe direction

Standard toe angle is recorded by the walkway as a signed value. We split the original toe angle value into two features: magnitude and direction. We keep the absolute value of deviation in toe angle and store the toe direction as a binary categorical feature, with 0 denoting negative toe angle and 1 denoting toe angles greater than or equal to zero

Hull area

To better approximate the actual shape of the footprint, we calculated the convex hull enclosing the point cloud for each footprint. The hull area is the enclosed area of the line segments bounding the footprint tightly in a convex hull. Figure 5A shows the hull area

BOS area

In gait, the BOS [23] is commonly measured as a one-dimensional length. We construct a convex polygon enclosing two footprints (a footprint and its preceding print) and then use this to approximate the area of the BOS. Figure 5B shows the picture of BOS area

LOP deviation angle

The deviation angle from an expected normal line of progression. A LOP is the line connecting the heel centers of two consecutive footfalls of the same foot. Ideally, the patient should be walking parallel with the walkway. We measure the deviation angle as the angle between:

- a line that starts on a foot and runs parallel to the walkway (the 'normal LOP')

- a line that starts at the center of the same foot, and stops at the center point of the next footfall of the same foot (the actual LOP)

Figure 5C shows the picture of line of progression deviation angle