Impairments of gait and balance are among the most incapacitating and least well-understood symptoms of Parkinson's disease (PD). In individuals suffering from such deficits, ambulation and everyday independence get severely affected due to the high risk of fall-related injuries. Well-established neuromodulation therapies addressing Basal Ganglia dysfunction in PD, which for decades have been optimized to address upper-limb motor symptoms, are highly effective for the symptomatic treatment of tremor, bradykinesia and rigidity. However they often fail to alleviate, or can even aggravate gait deficits. For instance, aspects such as gait initiation, postural instability and freezing of gait have been reported to worsen with continuous high-frequency (130Hz) deep brain stimulation (DBS) presumably due to the divergence in the nature and dynamics of the circuits that control leg function during gait. This has compelled the search for alternative ad-hoc non- continuous stimulation paradigms. Yet to date, the relevant approaches for therapeutic intervention remain unclear, and the underlying mechanisms largely unknown.
A key limitation holding back the design of evidence-based therapies is the lack of mechanistic readouts that correlate pathological neural activity patterns and leg dysfunction during gait. However this identification is contingent on technologies, concepts and methodologies allowing to simultaneously record and link brain states to whole-body biomechanical features representative of gait deficits.
To address this knowledge gap, project gaitDCODE established a high-resolution tracking platform at the Lausanne University Hospital, endowed with unique equipment for recording and modulating gait wirelessly and in real-time, in order to map the activity of subthalamic nucleus onto kinematic, kinetic, and muscle activity patterns while patients execute a range of activities of daily living. We further established the analytical concepts and methodologies necessary to identify and extract robust features representative of gait deficits from the high-dimensional, time-varying biomechanics that govern locomotion. Jointly, this approach allowed to record and link brain and locomotor states with a level of detail not yet shown in Parkinson's disease.
gaitDCODE followed an incremental roadmap, and identified neural correlates underlying leg motor function (and in particular of leg force modulation). It then leveraged this understanding to develop decoding algorithms that accurately predicted leg force modulations in real-time for different conditions, hence opening the possibility to design closed-loop neuromodulation therapies that may be automatically address such motor requirements based on online feedback of leg motor performance.
