Page 83 - Read Online

Nagwade et al. Soft Sci 2023;3:24 https://dx.doi.org/10.20517/ss.2023.12 Page 17 of 25

features from 180 sEMG signals from every subject undergo KNN and SVM techniques. After applying
these two techniques to the mined feature set, it was concluded that SVM showed a better overall accuracy
score of 97.3%, while KNN was 91.6%. This suggests that applications requiring hand gesture recognition
can utilize SVM classifiers as they have significant results in comparison to KNN.

Innovation is not only limited to the development of the interface but also its application. Dong et al.
developed a self-adhesive, semi-transparent dry electrode that is specifically used for lip-reading by utilizing
sEMG biopotential signals from the facial muscle movement . Machine learning algorithms are applied in
[120]
the processing, enabling the EMG signals to be converted into audible words, as shown in Figure 9D. Such
applications can be extremely useful in the medical and healthcare industry as it enhances patient-medic
communication. Moreover, the use of facial EMG can allow immersive interaction in a virtual world as well.

When direct comparisons are made to the standard Ag/AgCl electrode interface, the performance of these
soft biopotential sensors has shown comparable or even better results. For instance, low skin-electrode
contact impedance is an important factor for biopotential interfaces attached to the epidermal layer of the
skin for recording signals. The flexible sEMG electrode by Zeng et al. was fabricated using hydrographic
printing and showed an impedance value of around 10 KΩ at 500 Hz, whereas the traditional electrode
interface had a value of around 14 KΩ at 500 Hz. Moreover, at lower frequencies, the flexible interface
showed significantly lower impedance. Additionally, the SNR values Li et al. reported were comparable
between their semi-dry EEG interface and conventional ‘wet’ EEG electrodes. Both devices showed SNR
values of around 7 dB in the eyes open/close paradigm experiment. In other SSVEP paradigms, the semi-dry
electrode interface performed slightly better. The Gecko-inspired dry ECG electrode displayed its adhesion
capability by achieving over 30 cycles of adhesion to the skin. Moreover, the adhesion force was comparable
to those of wet adhesive, -1.3 N/cm2. Qualitative analysis is made between traditional and soft biopotential
interfaces concerning their wearability aspect, as shown in Table 1.

Neuromuscular stimulation with wearable non-invasive soft interfaces
Recording biopotential signals using skin-contact surface interfaces has a vast range of applications. Apart
from recording biopotential signals, there are applications where the stimulation of nerves is required for
rehabilitation purposes [121,122] . Non-invasive functional electrical stimulation (FES) can provide a pathway for
non-surgical solutions for brain and nerve stimulation. Moreover, such wearable electroceuticals can allow
for therapeutic electrostimulations without the limitations of invasive techniques.

Ohm et al. developed a soft hydrogel-based interface and performed neuromuscular electrical
[123]
stimulation . The group stimulated the tibialis anterior muscle of the leg and observed angular movement
at the ankle. The Ag-hydrogel interface performed better than the ionic hydrogel, as seen in Figure 10A.
Additionally, the performance of the Ag-hydrogel electrode interface was comparable to the commercial
electrodes. Similar experiments were repeated on the forearm muscles. With characteristics such as
stretchability, biocompatibility, and high electrical conductivity, such a soft muscle-stimulating interface has
the potential to replace conventional interfaces in the future.

[124]
Peripheral nerve stimulation (PNS) by using non-invasive methods has also been studied . By using high-
frequency sine-wave carriers, Botzanowski et al.. showed that it is possible to temporally interfere at deep
peripheral nerve targets. Temporal interference nerve stimulation (TINS) delivers effective outputs at a
lower current than standard transcutaneous electrical stimulation. As seen in Figure 10B, this flexible
PEDOT:PSS interface can non-invasively evoke a PNS response. Invasive methods are generally used to
achieve better selectivity, and therefore, implantable electrodes are favored for PNS. However, this non-