*2.1. Subjects*

Two subjects who had previously suffered from FNS during CI stimulation and had been re-implanted with Oticon Medical Neuro-Zti devices [8] were further investigated during a clinical follow-up visit. One of these patients was initially fitted with Advanced Bionics (Valencia, CA, USA) HiRes Ultra 3D implants and a mid-scala electrode in both ears. The other patient had originally been fitted with a MED-EL SYNCHRONY implant with a FLEX28 electrode in one ear.

Our goal was to gain deeper insight into the subjects' behavioral perceptions of different stimuli, with the ultimate aim of further improving their clinical outcomes.

#### *2.2. Stimuli*

We used the Oticon Medical Research Platform (OMRP) [9] to directly stimulate single electrodes on the Neuro-Zti Implant in two ways: (1) using stimulus parameters designed to replicate those of their prior CI, and (2) using the parameters of their current clinical CI mode. With Subject S2, we also investigated reversing the polarity of these pulses for a total of 4 stimulus types (Table 1 and Figure 1). All stimuli were presented in a pulse train with a 50% duty cycle (500 ms on and 500 ms off) using pulses presented at a rate of 500 Hz.


**Table 1.** Stimulus parameters that were used during this study.

<sup>1</sup> Subject S1 used duration coding for stimuli A− and B+ (current = 0.6 mA). Subject S2 used duration coding for stimuli B−/B+ (current = 0.4 mA) and current coding for stimuli A+/A− (duration = 30 μs). Note: We improved the software between S1 and S2 to allow current coding. <sup>2</sup> Distributed-all-polar (DAP) [7] indicates that current returns via the case electrode, like MP (monopolar) grounding, and simultaneously to all non-stimulating intra-cochlear electrodes, sometimes called 'common ground' (CG) (Figure 1).

**Figure 1.** Panel (**a**) shows a diagram of the Neuro-Zti implant illustrating current paths (dotted lines) for monopolar (MP) grounding used for stimuli A− and A+ (blue) vs. distributed-all-polar (DAP) grounding used for B+ and B− (orange). It also shows the electrode positions and numbering. Panel (**b**) shows sketches of the pulse shapes used in the experiment for MP grounding (blue) and DAP grounding (orange). The first row shows the clinical pulses in each case, with arrows denoting clinical loudness coding; the second row shows the reversed polarity pulses used with Subject S2.

In each trial, we gradually increased the stimulation charge (i.e., pulse duration or amplitude) until either (1) the sound was too loud or (2) non-auditory sensations became too intense for the subject.

#### *2.3. Procedure*

Testing was conducted during a scheduled clinical fitting session. The session was divided into three parts, each lasting about 30 min, interspersed with 5-min breaks. The total testing time was approximately 2 h.

In the first session, we selected suitable electrodes for the experiment from the regions where subjects had previously reported strong FNS responses (Figure 2). We conducted a search to determine which electrode to use for each subject utilizing the 'A−' stimulus (Table 1). This was chosen to mimic their previous cochlear implant and thereby give a good chance of eliciting an FNS response. For the search, we incrementally increased the pulse charge until either the subject reported a non-auditory sensation or the loudness reached an uncomfortable level. Not all electrodes that previously caused FNS (Figure 2) induced FNS responses with the Oticon Medical implant. Those that did, did so at different charge levels than previously reported (higher ones for subject S1, and lower ones for subject S2). In the end, we selected the electrode that elicited the most substantial FNS response to study in detail.

**Figure 2.** Selected results from case reports presented in [8]. C-level before and about half a year after re-implantation with Oticon Medical Neuro Zti EVO. Asterisks refer to the FNS thresholds with the previously implanted electrode. The upper x-axis refers to the channel number (#) of the previously implanted electrode; the lower x-axis refers to those of the EVO electrode. (**a**) Subject S1, right; (**b**) Subject S1, left; (**c**) Subject S2, right.

In the two following sessions, we stimulated the chosen electrode using either (1) the current clinical mode of their Oticon Medical CI (stimulus B+) or (2) a non-clinical mode designed to mimic their previous CI (stimulus A−). This allowed us to directly compare the in-situ effect of these stimulation modes while keeping implant hardware and subject factors consistent. For each stimulation mode, we gradually increased the intensity. At each charge level, we asked the subjects to: (1) rate the loudness on a scale from 0 (inaudible) to 10 (very loud); (2) describe any non-auditory sensations qualitatively; and (3) indicate whether they were comfortable enough to continue. Due to time constraints, we did not interleave testing modes; this would have necessitated reconfiguring the implant between presentations.

For subject S1, only stimuli of type A− and B+ were explored as described above, because it took some time until we found an electrode that exhibited strong non-auditory side effects. For all stimuli presented to subject S1, charge was increased using pulse duration due to a software limitation (Table 1). Fortunately, for subject S2, the software limitation was overcome, and we then used current coding for stimuli A−/A+ and duration coding for stimuli B−/B+ as would have been carried out clinically. We also had time to explore the effect of reversing pulse polarity for subject S2 in session 3 (Figure 1).

#### **3. Results**

The results of the main single electrode experiments for both subjects across all stimulus types are shown in Figure 3 as loudness growth curves. Reported non-auditory/FNS sensations are overlayed as separate symbols. For both subjects, the charge required to reach equivalent loudness levels was higher for stimulus B+ than for A− by more than a factor of 2. However, the growth in loudness was very different between the two. Subject S1 exhibited a rightward shift (~20 nC) and a slower loudness growth for the B+ stimulus, while Subject S2 only exhibited the rightward shift.

**Figure 3.** (**a**) Subject-reported loudness vs. individual pulse charge delivered for subject S1 (left electrode E5) showed faster loudness growth with stimulus A− than with B+. Stimulus A− also elicited non-auditory FNS sensations. (**b**) Subject S2 (right electrode E18) reported similar slopes but rightward-shifted loudness growth functions between A− and B+. Non-auditory sensations on both panels show how stimulus A− led to FNS stimulation at lower charge levels than for other stimulus types and how non-auditory sensations were affected in a markedly different way from the loudness percepts. Using stimulus B−, we were unable to achieve sufficient loudness. Instead, pronounced side effects were observed.

Reversing the polarity highlighted how non-auditory sensations were affected in a markedly different way than the loudness percept. More detailed results for each subject are presented individually below.
