Personalized Algorithms Enhance Walking in Parkinson’s Through Smarter Brain Stimulation

Doctors at the University of California, San Francisco (UCSF) have developed a personalized, data-driven method to improve walking in people with Parkinson’s disease. Their study used mathematical models paired with deep brain stimulation (DBS) to fine-tune settings for each patient. This approach helped participants walk faster and more steadily without worsening other symptoms . Parkinson’s often causes walking difficulties, including slow movement and imbalance, which raise the risk of falls and reduce independence. DBS is a surgical therapy where electrodes are implanted in brain regions involved in movement. It delivers electrical pulses to help restore more normal neural activity. Although DBS works well for tremor or stiffness, its effects on gait are inconsistent . In the UCSF study, three people with Parkinson’s who already had DBS implants participated. The implants included electrodes placed both deep in the brain (globus pallidus) and on the outer brain surface (motor cortex). These allowed researchers to record the electrical signals during normal walking and to analyze how different settings impacted gait performance . Using a “Walking Performance Index” (WPI)—a measure combining speed, stride stability, and other gait features—the team trained an AI model called a Gaussian Process Regressor. This model predicted the optimal DBS settings for each individual to improve WPI. These personalized settings led to significant improvements in walking, such as faster and more stable steps, without increasing other motor issues . The researchers also found that better walking was linked to lower beta-frequency brain activity in the globus pallidus during key phases of gait. This connection may serve as a neural biomarker for predicting which DBS settings best enhance walking . This study builds on broader findings that adaptive DBS—systems using AI to sense brain activity in real time and adjust stimulation—can halve motor symptoms compared to standard DBS in small clinical trials. Adaptive DBS has already shown promise in trials published in Nature Medicine and received FDA approval for investigational use . The UCSF work reinforces the potential of personalized, AI-guided brain stimulation to address one of Parkinson’s most problematic motor symptoms: gait dysfunction. By adjusting stimulation in a tailored way, this method could offer better outcomes than the one-size-fits-all DBS currently in wide use. Although based on only three participants, these results are encouraging. They point toward smarter DBS therapies that respond to each person’s unique brain signals. Further studies involving more participants will be needed to confirm safety and effectiveness and to help bring this technology into routine clinical use.