Applications of Artificial Intelligence in Movement Disorders, in the Pursuit of Personalized Healthcare

Artificial intelligence (AI) is increasingly finding applications in medicine. The access to detailed motor manifestations data (e.g., from wearables), electronic health records, research registries, and digital medical imaging in movement disorders (MD) makes them amenable to AI. This has the potential to tailor care to individual needs, thus providing personalized healthcare. AI has been studied in research, diagnosis, monitoring, and treatment, in MD. An ML algorithm has been used in a large national database to identify drugs with a potentially protective association with Parkinson's disease (PD). ML models applied to detailed clinical data from observational cohorts has allowed for a better understanding of disease progression in Huntington's disease (HD), which can be applied to clinical practice and research. The use of ML in analyzing data on daily movements (through sensors), gait, drawing of spirals, and keyboard typing has shown potential in diagnosis and monitoring of disease in PD. Automated analysis of functional magnetic resonance imaging (MRI) and electroencephalogram (EEG) can help in MD diagnosis and in predicting the risk of post-deep brain stimulation (DBS) cognitive impairment, respectively. Challenges with ML include the need for high-quality training data, risk of using non-representative data, overfitting, and limited reproducibility. While the maturity of AI in medicine is already a reality in some therapeutic areas, the uptake in MD is still limited. However, there is considerable promise for the future as some of the early explorative concepts, such as those described in this chapter, mature and consolidate their robustness and meet regulatory quality standards.