Mobile devices for the real-time detection of specific human motion disorders

We propose a wearable wireless sensing system for monitoring human motion disorders. The system is designed to be used at home or outdoor, as a mobile healthcare device assisting the person during the daily activity. It provides in real time an early detection of specific motion disorders at their outset, with excellent performance in terms of sensitivity and precision. The system is composed of inertial measurement units, a station for the real-time processing (smartphone/tablet/PC), dedicated algorithms and, eventually, a headphone for auditory feedback. An auditory feedback can warn the patient about particularly dangerous situations as the freezing of gait in the case of Parkinsonian patients, and timely provide rhythmic auditory stimulations to release the gait block. Two different hard and soft implementations of the system are discussed in this paper. The first has just one sensor in a headset. This solution features a fine detection of body motion and in particular of trunk oscillations, easy wearability, through auditory feedback. It is particularly compact and energy efficient, since no wired/wireless connection is required to give the audio-feedback (which reflects on the battery life). However, the headset suffers of the presence of a joint (the neck), which can hide important features of very disordered gaits, as in the case of the Parkinson's disease or other neurodegenerative diseases. The second implementation has two sensors on the shins. It allows fine detection of gait features, and guarantees the best performance presented in the literature to date in terms of sensitivity, specificity, precision, and accuracy in detection of the gait freezing. As a drawback, it requires an additional device in the ear for the audio-feedback, which implies higher power consumption respect to the headset device, for the wireless communication to the microphone. Different recognition algorithms were implemented in the same board, using fusion of raw signals from accelerometers and gyroscopes. The two solutions, their implementations, and experimental results will be discussed in detail, outlining strengths and deficiencies of the twos