Machine Learning Classification of Diadochokinesia for a Hemiplegic Ankle Foot Complex Pair

Diadochokinesia involves the ability to alternate the activation between an agonist and antagonist muscle pair. The status of diadochokinesia constitutes a common feature of a neurological examination. Additionally, hemiparesis demonstrates impairment of the hemiplegic affected limb by contrast to the unaffected limb in the context of diadochokinesia, which can be quantified through the smartphone as a functionally wearable and wireless gyroscope platform. The ankle foot complex represents a candidate for the evaluation of diadochokinesia disparity with respect to the ability to conduct dorsiflexion and plantar flexion. The ability to modify between dorsiflexion and plantar flexion is an inherent role for the function of the stance phase during gait. A smartphone with an armband can be conveniently mounted about the dorsum of the foot for the acquisition of a gyroscope signal for quantifying the dorsiflexion and plantar flexion of the ankle foot complex. Using post-processing through software automation the signal data can be consolidated to a feature set for the purpose of machine learning classification. The support vector machine has been selected as the appropriate machine learning algorithm. With the amalgamation of these available technologies considerable classification accuracy has been achieved for the distinction between a hemiplegic affected ankle and unaffected ankle based on the application of the support vector machine in conjunction with a smartphone for the acquisition of the gyroscope signal respective of a hemiplegic ankle foot complex pair.