Fleming, J. E. | bioRxiv Science

bioRxiv2025

A clinical grade neurostimulation implant for hierarchical control of physiological activity

Bioelectronic implants for neurostimulation aim to steer disordered neurophysiological processes back towards a healthy state. However, physiology is subject to biological rhythms, including the circadian rhythm and the sleep-wake cycle. These predictable rhythms affect disease symptomatology, biomarkers used in closed-loop therapies, and a physiological systems expected response to stimulation. Therefore, therapeutic devices should incorporate feedforward elements to align algorithm parameters with predictable changes in physiological state, as a parallel of physiological rheostatic control. Here we introduce the DyNeuMo-2c, the first clinical-grade implant capable of delivering closed-loop neurostimulation while flexibly changing its functional configuration according to time of day. The device can chronically measure brain activity and motion state to track potential biomarker patterns in natural, out-of-clinic settings, allowing identification and targeting of patient-specific chronotypes. The system implements a hierarchical control flow, with baseline therapy set by a circadian scheduler, and adaptive policies layered to take effect based on specific biomarkers indicating patient and disease state. Using a benchtop validation setup, we demonstrate that the system has the required capabilities for delivering time-contingent closed-loop therapy in two established clinical use cases: Parkinsons disease and epilepsy. Next, we deploy the system in vivo to deliver closed-loop deep brain stimulation in a healthy non-human primate model of vigilance, highlighting the importance of synchronisation between device operation and physiological state in various conditions (task performance, unconstrained behaviour, and sleep). Time-of-day-dependent adaptation of closed-loop stimulation enabled modulation of both vigilance and behaviour. Overall, the novel device architecture provides a proof-of-concept for delivering time-contingent therapy in chronic therapeutic settings where biological rhythms are of key importance.

neuroscience↗

bioRxiv2019

Simulation of closed-loop deep brain stimulation control schemes for suppression of pathological beta oscillations in Parkinson's disease

This study presents a computational model of closed-loop control of deep brain stimulation (DBS) for Parkinsons disease (PD) to investigate clinically-viable control schemes for suppressing pathological beta-band activity. Closed-loop DBS for PD has shown promising results in preliminary clinical studies and offers the potential to achieve better control of patient symptoms and side effects with lower power consumption than conventional open-loop DBS. However, extensive testing of algorithms in patients is difficult. The model presented provides a means to explore a range of control algorithms in silico and optimize control parameters before preclinical testing. The model incorporates (i) the extracellular DBS electric field, (ii) antidromic and orthodromic activation of STN afferent fibers, (iii) the LFP detected at non-stimulating contacts on the DBS electrode and (iv) temporal variation of network beta-band activity within the thalamo-cortico-basal ganglia loop. The performance of on-off and dual-threshold controllers for suppressing beta-band activity by modulating the DBS amplitude were first verified, showing levels of beta suppression and reductions in power consumption comparable with previous clinical studies. Proportional (P) and proportional-integral (PI) closed-loop controllers for amplitude and frequency modulation were then investigated. A simple tuning rule was derived for selecting effective PI controller parameters to target long duration beta bursts while respecting clinical constraints that limit the rate of change of stimulation parameters. Of the controllers tested, PI controllers displayed superior performance for regulating network beta-band activity whilst accounting for clinical considerations. Proportional controllers resulted in undesirable rapid fluctuations of the DBS parameters which may exceed clinically tolerable rate limits. Overall, the PI controller for modulating DBS frequency performed best, reducing the mean error by 83% compared to DBS off and the mean power consumed to 25% of that utilized by open-loop DBS. The network model presented captures sufficient physiological detail to act as a surrogate for preclinical testing of closed-loop DBS algorithms using a clinically accessible biomarker, providing a first step for deriving and testing novel, clinically-suitable closed-loop DBS controllers.

bioengineering↗