Analisi approfondita del fenomeno di freezing of gait con l’utilizzo di sensori inerziali: risposta alla levodopa e agli eventi trigger = A deeper insight into the freezing of gait phenomenon using wearable sensors: response to levodopa and triggering events

Parkinson's disease is the second most common neurological illness, associated with early prominent death of dopaminergic neurons leading to a movement disorder with classical parkinsonian motor features and further non-motor complications. Freezing of gait is among the most disabling, complex and less known symptoms, in terms of clinical expression, pathogenesis and treatment. It is expressed as sudden and brief episodes of inability to produce effective stepping, as if feet are glued to the floor. Consequently, freezing is a major concern in Parkinson’s disease management due to its link to falls and poor quality of life. Nevertheless, it is surprisingly difficult to assess in daily clinical practice, as it is mainly self-reported by patients or caregivers and particularly troublesome to be elicited and estimated in hospital settings. For this reason, freezing of gait severity cannot be routinely monitored but only described in general terms that are not sensitive to changes ensuing with disease progression. Recently, wearable sensors have been employed in order to quantitatively and objectively assess this symptom. Lately, some papers have also been published in the direction of its prediction. On the other hand, a better understanding of the phenomenon is highly needed to aid the development of effective, personalised therapeutic strategies. The aim of the present study is to provide a deeper insight into the motor manifestations of freezing, in order to gather precise information about its frequency, duration, triggering circumstances and responsiveness to levodopa. For this purpose, a simple, small and low cost system made of two inertial measurement units placed on each thigh was exploited. A total number of 23 participants took part in the experimental protocol, including dopaminergic-responsive freezers (30.4%), dopaminergic-induced freezers (17.4%) and dopaminergic-resistant freezers (52.2%): 95 freezing events were recognised during the procedure. The latter consisted of a Timed-Up and Go test performed twice (with a right/left U turn) in both therapeutic conditions. Signals processing of angular velocity and acceleration was executed offline and freezing episodes were discriminated from normal gait. Freezing events were further differentiated according to their eliciting cause, as reported in Fahn’s classification: start hesitation, apparent hesitation in tight quarters, turn hesitation and destination hesitation. From each signal segment related to freezing of gait, the power spectral density was computed in order to perform frequency analysis between triggering events and pharmacological states. The dynamic time warping technique was implemented to detect statistically significant dissimilarities between different phenotypes. Features extraction was carried out and 192 parameters, both in time and frequency domain, were investigated in order to discern between the symptom subtypes. Results were promising: wearable sensors allowed to underline the existence of distinctive factors, thus corroborating the hypothesis that several clinically similar but physiologically distinct phenomena could be pulled together under this motor symptom. Although the size of the dataset is not sufficient to validate the outcomes, its wide heterogeneity is unusual in similar studies. Thus, it suggests the presence of more than one freezing of gait subtype, each with a specific response to dopaminergic administration.