Closed-loop algorithm guided programming with multiple inputs for optimization of Deep Brain Stimulation settings

TY - CHAP

T1 - Closed-loop algorithm guided programming with multiple inputs for optimization of Deep Brain Stimulation settings

AU - Gülke, Eileen

AU - Domschikowski, Mirjam

AU - Juarez-Paz, Leon

AU - Scholtes, Heleen

AU - Paschen, Steffen

AU - Pötter-Nerger, Monika

PY - 2022

Y1 - 2022

N2 - Keywords:Deep Brain Stimulation, Closed Loop Algorithms, Semi-Automatic Programming Objectives:Evaluate the feasibility of a semi-automatic algorithm-guided-programming (AgP) strategy for Parkinson’s disease subjects with bilateral Nucleus subthalamicus (STN) directional Deep Brain Stimulation (DBS) to reduce the programming effort for beneficial DBS settings. Methods:The AgP iteratively suggests new DBS settings based on clinical responses to previously tested DBS settings until it converges. The algorithm input is calculated at each iteration by the weighted combination of multiple motor symptom scores, whose weights are determined based on their response to either Levodopa or DBS. 20 PD subjects were enrolled at two German sites as part of the CLOVER-DBS study. Subjects had stable DBS settings for at least 4 weeks following Standard of Care (SoC) programming. After 12 hours of antiparkinsonian medication withdrawal (MED-OFF condition), motor symptom severity was assessed by Unified Parkinson’s disease rating scale (UPDRS) III scores and using a sensor-based system (Kinesia™ One) in the SoC DBS-ON and DBS-OFF conditions. For each hemisphere, DBS settings optimization by the AgP was performed using manually assessed and sensor-based symptom scores. Programming effort in terms of time and explored DBS settings was quantified for AgP, and the clinical efficacy of SoC and AgP DBS settings was compared in a randomized, crossover double-blind fashion by UPDRS III scores. Results:UPDRS III scores significantly (p < 0.001) improved 63.2 (42.6 | 71.9)% for AgP and 55.3 (36.9 | 67.5)% for SoC DBS settings when compared to MED-OFF/DBS-OFF scores. UPDRS III scores differences for AgP and SoC DBS settings were not significant. Per subject, AgP tested 35.0 (28.5 | 37.0) DBS settings before the algorithm converged, resulting in programming times of 1.7 (1.6 | 2.0) hours. However, AgP found optimal DBS settings after testing only 21.0 (15.5 | 30.0) DBS settings, which could result in programming times of 1.0 (0.7 | 1.7) hours. Conclusions:AgP proved feasible and efficient for finding optimal acute DBS settings. Long-term clinical results of AgP and its potential to reduce programming effort still need to be investigated.

AB - Keywords:Deep Brain Stimulation, Closed Loop Algorithms, Semi-Automatic Programming Objectives:Evaluate the feasibility of a semi-automatic algorithm-guided-programming (AgP) strategy for Parkinson’s disease subjects with bilateral Nucleus subthalamicus (STN) directional Deep Brain Stimulation (DBS) to reduce the programming effort for beneficial DBS settings. Methods:The AgP iteratively suggests new DBS settings based on clinical responses to previously tested DBS settings until it converges. The algorithm input is calculated at each iteration by the weighted combination of multiple motor symptom scores, whose weights are determined based on their response to either Levodopa or DBS. 20 PD subjects were enrolled at two German sites as part of the CLOVER-DBS study. Subjects had stable DBS settings for at least 4 weeks following Standard of Care (SoC) programming. After 12 hours of antiparkinsonian medication withdrawal (MED-OFF condition), motor symptom severity was assessed by Unified Parkinson’s disease rating scale (UPDRS) III scores and using a sensor-based system (Kinesia™ One) in the SoC DBS-ON and DBS-OFF conditions. For each hemisphere, DBS settings optimization by the AgP was performed using manually assessed and sensor-based symptom scores. Programming effort in terms of time and explored DBS settings was quantified for AgP, and the clinical efficacy of SoC and AgP DBS settings was compared in a randomized, crossover double-blind fashion by UPDRS III scores. Results:UPDRS III scores significantly (p < 0.001) improved 63.2 (42.6 | 71.9)% for AgP and 55.3 (36.9 | 67.5)% for SoC DBS settings when compared to MED-OFF/DBS-OFF scores. UPDRS III scores differences for AgP and SoC DBS settings were not significant. Per subject, AgP tested 35.0 (28.5 | 37.0) DBS settings before the algorithm converged, resulting in programming times of 1.7 (1.6 | 2.0) hours. However, AgP found optimal DBS settings after testing only 21.0 (15.5 | 30.0) DBS settings, which could result in programming times of 1.0 (0.7 | 1.7) hours. Conclusions:AgP proved feasible and efficient for finding optimal acute DBS settings. Long-term clinical results of AgP and its potential to reduce programming effort still need to be investigated.

U2 - https://doi.org/10.1159/000524634

DO - https://doi.org/10.1159/000524634

M3 - Konferenzbeitrag - Poster

BT - WSSFN 2022

ER -