Deep brain stimulation (DBS) remains one of the most effective advanced therapies for Parkinson’s disease (PD), yet considerable heterogeneity persists in clinical outcomes, patient selection, and long-term neuropsychiatric effects. This narrative review synthesizes recent evidence on the mechanisms, efficacy, and evolving applications of DBS, integrating findings from neurophysiology, neuroimaging, genetics, and computational modeling. Current data indicate that stimulation of the subthalamic nucleus (STN) and globus pallidus internus (GPi) reliably improves cardinal motor symptoms and reduces dopaminergic medication requirements, with sustained benefits over time. However, axial symptoms, including gait disturbance, postural instability, and speech, demonstrate variable or limited responsiveness, underscoring the complexity of the underlying neural circuits. Non-motor outcomes also show substantial variability: sleep and gastrointestinal functions often improve, whereas apathy, impulsivity, and speech disturbances may emerge or worsen in selected patients. Emerging biomarkers, including evoked cortical responses, oscillatory signatures, radiomic features, and gray matter morphometry, show promise in refining patient-specific predictions, although external validation remains limited. Genetic variants, particularly in GBA and LRRK2, further modulate cognitive trajectories and long-term outcomes. Advances in imaging-guided “asleep” DBS, directional leads, and personalized programming strategies have improved precision and safety. Concurrently, machine learning and adaptive DBS paradigms represent major future directions, enabling real-time, biomarker-driven modulation tailored to disease dynamics. By integrating multidomain evidence, this review highlights both the established therapeutic value and the persistent challenges of DBS, emphasizing the need for mechanistically informed, precision-based approaches. These developments collectively outline a path toward more individualized and circuit- specific neuromodulation strategies in PD.
