Matters arising: bridging predictive algorithms and clinical practice in DBS contact selection

Optimal contact selection is a critical determinant of deep brain stimulation (DBS) outcome in Parkinson’s disease, yet it remains one of the most challenging steps in clinical practice. When the optimal stimulation contact is identified directly, precise electrical parameters can be applied without exhaustive testing at every individual electrode contact, saving clinical time and reducing patient burden 1 . Conversely, suboptimal contact selection carries consequences that extend well beyond inefficiency: when the therapeutic target is missed, side effects become more likely. In search of adequate effect, clinicians are then compelled to explore a wide range of contacts and parameter combinations, exposing patients to repeated adjustments, conflicting sensations, and the confusion and overload that comes with prolonged and inconclusive programming trajectories. Neural biomarkers, specifically local field potentials (LFPs) recorded from chronically implanted DBS devices, offer a promising route toward network-guided, patient-specific programming. By revealing the spatial distribution of pathological beta band ( ± 13–35 Hz) activity, LFPs could enable direct identification of the optimal contact and, ultimately, support adaptive DBS systems that continuously adjust stimulation to the patient’s momentary neural state. Realizing this potential, however, requires that LFP-based contact selection algorithms are not only accurate but robust and reproducible across clinicians and centers, a standard that has yet to be rigorously established.