**1. Introduction**

Essential tremor (ET) is the most prevalent movement disorder, a ffecting up to 4% of the adult population [1]. Medical managemen<sup>t</sup> of ET is limited and is often unsatisfactory [2]. For medically refractory cases, deep brain stimulation (DBS) may be considered. The long-term safety and e fficacy of DBS are well-established [3], although it is not certain whether its initially reported superior tremor suppression is also achieved in successive cohorts [4]. Additionally, evidence shows that 33 out of 45 patients in one study (73.3%) reported waning tremor control at a mean time of 18.8 ± 15.1 months postoperatively [5].

The outcome of ET-DBS depends on several factors. Preoperative considerations include but are not limited to tremor characteristics [6] and the anatomical target of the intervention [7,8]. Postoperatively, optimal tremor reduction is achieved with time-consuming programming. The current strategy for symptom control starts with a standardized evaluation of several conventional stimulation parameters, representing the highest probability for success. Pulse widths with an estimated chronaxie (i.e., the minimum time for exciting a neural element using half the intensity to elicit a threshold response, for review see [9]) for myelinated axons in ET-DBS average 40–90 μs [10]. As far as frequency is concerned, it should be noted that during the first application of DBS, 50 Hz stimulation was considered as high-frequency stimulation [11]. Ever since, a broad range of stimulation frequencies up to 185 Hz has been explored, although there is no clear relation between the stimulation frequency and degree of tremor suppression. All parameters are further titrated in a 'trial and error' fashion, until satisfactory tremor suppression is achieved in the absence of side effects. In practice, this requires extensive programming sessions, in which patient fatigue may hamper achieving the desired results. Empirical titration becomes additionally challenging when conventional DBS parameters do not address individual requirements [12] or become subject to tremor habituation [13], requiring broader parameter searches than feasible. A clear understanding of the therapeutic role of the stimulation parameters is therefore essential. Unfortunately, the relation of any of the stimulation parameters and degree of tremor suppression remains insufficiently understood. A study exploring high-frequency stimulation as a putative cause for worsening balance in ET patients demonstrated that reducing stimulation frequency from 170–185 to 130 Hz after optimizing tremor control improved axial cerebellar signs [14]. In addition, therapeutic DBS intensity levels suppress tremor, while supra-therapeutic amplitudes and pulse-widths cause (gait) ataxia [15,16]. The deleterious effect of excessive stimulation translates thus into a narrow therapeutic window for tremor suppression.

Despite efforts to individualize and improve DBS programming [17,18], these are either not robust enough or are too technically challenging to be routinely applied in clinical practice. In the absence of an explicit, validated programming protocol, the process remains laborious and outcomes inconsistent, with average tremor reduction varying between 33 and 74% [2]. Outcomes may be improved with expert programming, shown to provide significant improvement in 37% of patients and partial improvement in 15% [12].

In this study, we aim to improve DBS programming protocols. We start by reviewing all documented DBS parameters and tremor outcomes in ET, with the aim to gain insight into the advancement of DBS programming over time, as well as to characterize the role of the DBS-parameters (voltage, pulse width, and frequency). Next, as a proof of concept, we introduce a novel approach for a timely and thorough exploration of the DBS parameter space in individual patients.
