Abstract
Many recent studies have highlighted that the harmony of physiological walking is based on a specific proportion between the durations of the phases of the gait cycle. When this proportion is close to the so-called golden ratio (about 1.618), the gait cycle assumes an autosimilar fractal structure. In stroke patients this harmony is altered, but it is unclear which factor is associated with the ratios between gait phases because these relationships are probably not linear. We used an artificial neural network to determine the weights associable to each factor for determining the ratio between gait phases and hence the harmony of walking. As expected, the gait ratio obtained as the ratio between stride duration and stance duration was found to be associated with walking speed and stride length, but also with hip muscle forces. These muscles could be important for exploiting the recovery of energy typical of the pendular mechanism of walking. Our study also highlighted that the results of an artificial neural network should be associated with a reliability analysis, being a non-deterministic approach. A good level of reliability was found for the findings of our study.
Highlights
Strokes are one of the most important leading causes of movement disability in Europe and the US; each year approximately 780,000 people experience a new (77%) or recurrent (13%) stroke, with 30–35 deaths for every 100,000 people [1]
Instrumented gait analysis, based on stereophotogrammetry, force platforms, electromyography, and occasionally metabolimeters and/or inertial wearable devices can allow the characterization of stroke gait patterns through a huge quantity of kinematic and kinetic data, which can be added to clinical data [4,5,6]
The problem is solved by a cut dividing the segment in a proportion equal to 61.8%–38.2%, which is very close to the division in stance and swing phases of the physiological gait cycle in healthy subjects
Summary
Strokes are one of the most important leading causes of movement disability in Europe and the US; each year approximately 780,000 people experience a new (77%) or recurrent (13%) stroke, with 30–35 deaths for every 100,000 people [1]. Artificial neural networks have been used as models for screening [8], risk identification [9] or as a prognostic tool [10], especially when the possible relationships among factors are not linear This is the case for harmony in walking, a feature that can be assessed, according to suggestions, by the ratio between stride and stance [11]. Scheffer and Cloete had the intuition of the potentialities of combining two emerging technologies: artificial neural networks and motion capture [16] Their results suggested the usability of ANN and gait analysis for planning gait rehabilitation therapy and monitoring its outcomes in the case of a stroke
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
