The use of digital biomarkers, advanced technology in DBS programming and machine learning to diagnose dystonia and other movement disorders is growing exponentially.
Digital Biomarkers
Digital biomarkers have been tried to assess various motor symptoms of Parkinson’s disease through numerous means including wearables, use of smartphone applications, as well as more recently remote sensing devices. While the studies have been small, they have been promising in terms of providing ecological validity of patient motor symptoms, filling in gaps of unrecognized motor symptoms, and improving medication adherence.
However, their use for non-motor symptoms is limited and there are still barriers to scalability in terms of cost, privacy concerns, lack of supportive infrastructure for patients and clinicians, and lack of targeted data for clinicians to provide meaningful clinical recommendations.
DBS Programming for Parkinson’s
Regarding deep brain stimulation programming for Parkinson’s disease, technological advancements have been moving along in the right direction. While the traditional methods are time-consuming, labor-intensive, require multiple trials and errors, and lack precision and flexibility, advanced software offers more intuitive interfaces for enhancing the programming.
MRI-guided programming can facilitate faster programming via direct visualization of intended targets. Sensing technology provides precise symptom-specific biomarker-guided therapy. At-home sensing allows several opportunities, including remote monitoring of disease state, symptom fluctuations, better assessment of medication off and on state, and tracking medication compliance. Directional leads allow the steering of the electrical fields to target specific brain regions, and using short pulse widths can avoid unnecessary stimulation of large fiber pyramidal tracts.
Indeed, teleprogramming offered by all DBS platforms is convenient and safe. The ongoing research in adaptive DBS technology will foster optimal energy-efficient programming.
Finally, AI-based automated programming protocols have the potential to revolutionize DBS therapy by improving efficiency, precision, personalization, and adaptability while reducing the burden on clinicians.
Deep Learning in Dystonia
Recent years have seen rapid progress in understanding the pathophysiology of dystonia, which, collectively, shifted its definition toward a neural network disorder. In parallel, we have witnessed an exponential growth of various applications of artificial intelligence (AI) methodologies, from our daily lives to healthcare.
Rethinking ways for clinical management of patients with dystonia, a few studies have explored the power of analytical AI for predictive diagnosis and therapeutic outcomes. Currently, the diagnosis of dystonia is largely formulated on clinical syndrome characteristics, without objective biomarkers or gold standard tests, leading to an average of 10-year delay in confirmed diagnosis. In turn, delays in diagnosis lead to deferred treatment of these patients.
Recent studies have demonstrated that deep learning platforms, named DystoniaNet and DystoniaBoTXNet, are capable of fully automated discovery of an objective neural network biomarker from a single structural MRI of a patient with focal dystonia for objective, accurate, and fast diagnosis of the disorder and predictive outcome of botulinum toxin treatment.
Given these encouraging findings, ongoing clinical trials are testing the translational potential of these platforms for their integration into the clinical management of patients with dystonia.