Modeling the human gait phases using granular computing
Gait analysis is applied for the provision of diagnosis, evaluation, and for the design of therapeutic intervention for subjects suffering from neurological disorders. The benefits accruing from gait analysis are well established. People with neurological disorders like mild traumatic brain injury, Cerebral Palsy and Multiple Sclerosis, suffer associated functional gait problems. The symptoms and sign of these gait deficits are different from subject to subject and even for the same subject at different stage of the disease. Identifying these gait related abnormalities helps in the treatment planning and rehabilitation process. The dynamic behavior of gait parameters is cyclic and the "normal" or expected pattern or values of these parameters over a gait cycle is well known. Modeling dynamic gait parameters over a gait cycle helps identify alteration or deviation from the expected reference pattern or values. Specifically, quantifying the kinematics, the kinetics and the surface electromyography gait parameters over a given gait cycle play a crucial role in recognizing associated neurologically related gait deficits. Additional quantification and representation into the seven gait phases adds more reliability and specificity to the analysis process. The current gait assessment methods do not provide very specific information within the seven gait phases. Most gait modeling techniques are limited to full gait cycle analysis techniques that focus on comparison of reference patterns or values with the respective parameters of neurological impaired subjects. Attempts have been already made to model and represent gait parameters for each phase by averaging the sample values in the respective phase. Mean value representation may be a good way and works well for slowly time-varying signals. However, averaging is not a good choice for non-smoothly time-varying signals with typical peaks and valleys. Ground reaction forces and muscle activity signal are examples of such rapidly time-varying signals with characteristic shape and peak amplitudes. We believe that a more accurate modeling, representation, and quantification on each phase could be accomplished by employing granular computing scheme. Modeling gait parameter value in each phase based on data-driven granule representation helps to capture the signal information in each sub-cycle and preserve the experimental significance and justifiability of the signal. We present a novel granular based model of gait analysis that objectively quantifies and provides individual assessment and evaluation information within each sub-cycle. Each gait phase is treated as separate entity or as an information granule. Granule parameters are optimally determined from the measured signal samples in each gait phase. The quantitative measures would provide individual based impairment level information for each of seven gat phases. An appropriate similarity measure algorithm may be applied to compare the able-bodied group (reference pattern) with the impaired subject (input pattern). Thus we design and implement new approach to detect gait variability after mild traumatic brain injury under the dual-task gait protocol. The new technique can also be used to diagnose mild traumatic brain injury, particularly in sports and in the military. Further, the granule representation was used to measure gait deficits in Multiple Sclerosis and Cerebral Palsy subjects. Our approach can help clinician measure gait deficits, and prescribe the right treatment and rehabilitation procedures.
Bogale, Melaku Ayenew, "Modeling the human gait phases using granular computing" (2013). ETD Collection for University of Texas, El Paso. AAI3565896.