Investigating the effects of phantom sensation on gait autocorrelation in transtibial amputees using signal processing techniques: A multidisciplinary analysis

Objective: Phantom sensation (PS) is defined as the sensation of the amputated limb that persists after amputation. The aim of this study was to investigate the effects of PS on the autocorrelation feature of gait. The information obtained will contribute to the methods of analysis and intervention in the rehabilitation. Methods: Unilateral traumatic trans-tibial amputees with phantom sensation (PG) (n = 10) and with no phantom sensation (N-PG) (n = 11), and healthy controls (HC) (n = 11) were recruited. Individuals walked on the treadmill at their preferred speed while inertial motion units recorded 512 consecutive steps. Gait analysis was repeated with a 5 % perturbation on the treadmill. The following signal processing methods were utilized to investigate autocorrelation function of gait: multiscale entropy (MSE), detrended fluctuation analysis (DFA), rescaled range analysis (RRA), autocorrelation function (ACF), power spectral density (PSD), Lempel Ziv complexity (LZC), Higuchi fractal dimension (HFD), Katz fractal dimension (KFD), Renyi entropy (RE), and Tsallis entropy (TE). Results: PG (p = 0.049; d = 0.905) had better gait autocorrelation results than N-PG (p = 0.005; d = 1.234) on both regular and perturbed ground compared to the HC. PG showed more similar gait adaptations to healthy individuals. The LZC criterion and HFD yielded more significant results in differentiating between the groups in both ground changes and existing ground conditions (p = 0.05). Conclusion: The results demonstrated that phantom sensation is a functional sensation that enhances the autocorrelation function of gait and contributes to the neuromotor nature of gait. The proposed signal processing techniques may become an alternative option to the methods previously used in this field.