Wearable body and wireless inertial sensors for machine learning classification of gait for people with Friedreich's ataxia

Abstract

The integration of wearable and wireless inertial body sensors with machine learning offers the capacity to diagnose neurological disorders involving gait. Clinical rating scales may be unable to offer precise measurement of gait dysfunction in Friedreich's ataxia compared to wearable body and inertial sensors. Using wireless inertial sensors mounted about the ankle joint of a person with Friedreich's ataxia, the accelerometer and gyroscope signal recordings can be wirelessly transmitted to a cloud computing resource for postprocessing, such as the development of a machine learning feature set. Machine learning can be applied to distinguish between the gait features of a person with Friedreich's ataxia and a person with healthy gait characteristics as a comparator through the application of a multilayer perceptron neural network. A considerable degree of classification accuracy for distinguishing between the gait feature set for the person with Friedreich's ataxia and healthy subject was achieved. The synthesis of wearable and wireless inertial body sensors with machine learning may offer the potential to enhance clinical diagnostic acuity and conceivably prognostic foresight.