Brain–Motor–Nutrition Digital Phenotyping in Parkinsonian Disorders: Linking Distributed Neural Dysfunction to Motor Biomarkers and Precision Intervention (Preprint)

<sec> <title>BACKGROUND</title> <p>Parkinsonian motor dysfunction is increasingly understood as a manifestation of distributed abnormalities involving cortical, basal ganglia, cerebellar, brainstem, sensory-integration, and executive-control networks. Conventional clinical assessments are episodic and may not adequately capture fluctuating or context-dependent motor abnormalities occurring during daily life. Mobile and wearable technologies provide an opportunity to continuously characterize these changes through digital biomarkers.</p> </sec> <sec> <title>OBJECTIVE</title> <p>This Review proposes a brain–motor–nutrition digital phenotyping framework that links distributed neural-network dysfunction to motor phenotypes, digital biomarkers, multimodal artificial intelligence (AI), and precision intervention. We particularly examine how mobile, wearable, and camera-based technologies can translate neurobiological abnormalities into longitudinal measures of functional motor performance.</p> </sec> <sec> <title>METHODS</title> <p>We synthesized evidence concerning neuroimaging, markerless three-dimensional (3D) human pose estimation, wearable sensing, digital biomarkers, multimodal AI, nutrition, and lifestyle-related modifiers in Parkinsonian disorders. Particular emphasis was placed on freezing of gait (FOG) as a model phenotype through which distributed neural dysfunction can be linked to objectively measurable motor behavior. The evidence was organized according to the conceptual pathway of brain-network dysfunction, motor phenotype, digital biomarker, and precision intervention.</p> </sec> <sec> <title>RESULTS</title> <p>Neuroimaging studies indicate that Parkinsonian motor dysfunction involves distributed abnormalities in functional connectivity, white-matter integrity, cerebral perfusion, dopaminergic signaling, and network organization. Markerless 3D pose estimation can quantify gait initiation, stride variability, turning dynamics, arm swing, postural control, transitional mobility, and whole-body coordination, while wearable sensors extend assessment into home and community environments. Multimodal AI may integrate these neural, behavioral, clinical, and physiological data streams to identify individualized motor phenotypes that are incompletely captured by conventional clinical scales. Nutritional status, sarcopenia, frailty, body composition, physical activity, sleep, and metabolic health may further modify motor performance, physiological reserve, and rehabilitation responsiveness.</p> </sec> <sec> <title>CONCLUSIONS</title> <p>Brain–motor–nutrition digital phenotyping provides a conceptual framework for connecting neural-network dysfunction with continuously measurable motor behavior and individualized intervention. Its translational potential includes remote monitoring of FOG and fall risk, longitudinal assessment of disease progression, treatment-response monitoring, rehabilitation personalization, and neuromodulation optimization. Prospective longitudinal cohorts, standardized digital motor endpoints, multicenter external validation, interpretable AI, harmonized medication and contextual states, representative populations, privacy-preserving data governance, and integration with clinical workflows are required before these approaches can achieve routine clinical utility. Mobile and wearable digital phenotyping may ultimately enable a shift from episodic Parkinsonian assessment toward continuous, mechanism-informed, and individualized monitoring.</p> </sec>