Development of a new high sensitivity mechanical switch for augmentative and alternative communication access in people with amyotrophic lateral sclerosis

The Lever Magnetic-spring Mechanical Switch (LeMMS) was specifically designed to allow activation with a very low contraction force and to provide a return force suitable to move the finger against gravity (to not require a finger extension for switch deactivation). It consists of two rectangular floating half-shells joined by a pin (with fulcrum function) at one end. A mechanical micro-switch (Huaweite Electronics Co. Ltd., mod. WK1–01), with a buckling-spring switch design, is activated by the slight rotation of one half-shell with respect to the other. A couple of magnets (Calamit© - Ferrite Cuboid Magnet, dimensions 9x5x5mm, adhesion 350 g. on 9x5mm² surface) facing each other with the same polarity works as an additional spring providing the repulsive force required to maintain the switch open in resting conditions. Hence, the functioning of LeMMS is based on the equilibrium relationship of the moment of forces reported below. Figure 1 shows the forces involved during use of the device.

In order to activate the switch, the moment of the force exerted by the finger should be higher than the sum of the forces exerted by the return spring of the micro-switch and by the magnets.

The following formula represents the equilibrium relationship: where:

F_F is the finger exerted force, X_F is the distance of F_F from the rotation axis, F_M is the repulsive force between magnets, X_M is the distance of F_M from the rotation axis, F_SW is the force exerted by the micro-switch spring and X_SW is the distance of F_SW from the rotation axis.

The finger-exerted force consists of two components, i.e.:

F_F = F_WF + F_P (where)

F_WF is the weight of the finger and F_P is the force actively exerted by the patient’s muscle contraction. Then from eq. (1) we obtain:

If as a first approximation we consider F_M, X_M, F_SW, and X_SW as constants that depend on the construction features of the device, it results that the force required to activate the device is a function of the finger position on its surface.

Looking at Fig. 2, one can see that a simple longitudinal sliding of the finger allows to find, in relation to the finger weight, the most appropriate distance from the fulcrum so that only a very small active force is required to activate the device (as little as a few grams).

Optimum force characteristics of a key require a steadily increasing force as it is depressed until contact is made. Immediately beyond the switch-contact point, the force is substantially changed so that users can “feel” when the key has been pressed [16].

Actually, the micro-switch spring force and the magnets’ repulsive force are not constants, but both depend on the rotation of the finger support plane. For very small rotation angles, the rotation can be approximated to a linear displacement (d) along the vertical axis (see Fig. 1).

The relationship between the key force and displacement of the micro-switch was assumed to be in accordance with the specifications reported in the ISO/DIS 9241 standard [17].

Further, the magnetic repulsive force depends on the distance squared for large displacements; in our case, the initial distance between the magnets is about 2.5 mm, which is reduced by 1 mm when pressed. Then, the repulsive force can be considered in a first approximation as linearly dependent on the displacement.

In order to provide a reliable model of the device we measured the actual characteristics of these force components using a simple experimental set-up including a micrometric surface gauge (NIIGATASEIKI VHK-15), and a load cell and bridge amplifier (HBM S2M, MVD 2420) [17, 18].

Figure 3 reports findings on the displacement-force relationship in a sample device. Repeated measurements on different component samples (same model #) revealed non-significant variability for the switch and magnets performance.

Both components can be modeled by a combination of linear elements and, then, eq. (2) is replaced by the following:

Key-press condition:

Release condition:

where:

K_M and F_M0 are respectively the slope and intercept of the line modeling the linear magnetic repulsive force, and K_SW, F_SW0 and F_SW1 are respectively the slope and the intercepts of the lines modeling the micro-switch spring forces during the key-press and release phases. It is worth noting that the linear model assumes the same slope for both phases and the presence of a hysteresis phenomenon.

All the elements of equations (3) and (4) except for X_F, d, F_P can be considered as constants that depend on the construction features of the device.

All variables were expressed as mean ± standard deviation (SD). The number of activations of LeMMS during the Click-Test-30 was compared across the four time-points (t0, t1, t2, t3) by Friedman’s test. Paired comparisons were analyzed by the post hoc Wilcoxon test and the Bonferroni correction for multiple comparisons was applied (statistical significance being set at p < 0.008). Sensitivity and specificity of the Click-Test-30 as a predictor of the ability to use the LeMMS for scanning access was assessed by means of receiver operating characteristic (ROC) curve analysis [23]. The optimal cut off score was chosen as the point that jointly maximized sensitivity, specificity and ability in classified patients. In addition, the area under the ROC curve (AUC) was calculated as an indicator of the overall ability of the Click-Test-30 to predict an efficient use of the scanning access in our patients. The greater the AUC, the greater the Click-Test-30’s ability to distinguish patients able vs. unable to access the scanning system using the LeMMS. An AUC value > 0.8 was classified as an excellent discrimination power indicator. In order to assess the sensor’s limit of applicability, ROC analysis was conducted on Click-Test-30 and muscle strength values. The optimal cut-off point of ROC curves was estimated using the method proposed by Froud & Abel [24], which indicates the point on the ROC curve closest to the top-left corner of ROC space. Consequently, we selected the point in the ROC space that minimized the quantity q defined as: q = min {(1 – sens)2 + (1 – spec)2} [24]. The calculation was made using the STATA command “rocmic” followed by the “bootstrap” command in order to generate the confidence interval. Statistical analysis was performed using the STATA 13.0 (Stata Corp LLC, College Station, Texas, USA) software package.

We found a substantial correlation between muscle strength and the Click-Test-30, Rho spearman = 0.75 (95% C.I. = 0.47–0.90, p = 0.0001). The ROC curve analysis reported in Fig. 5 shows ROC curves of the Click-Test-30 considering the type of communicator used at t0 and at 1-year follow-up. The area under the ROC curves (95% C.I.) was 0.99 (95% C.I. = 0.96–1.00) and 0.77 (95% C.I. = 0.51–0.91), respectively, for current ability and ability at 1-year to use a communicator. The Click-Test-30 was able to differentiate patients who were able to use their scanning access system through the LeMMS from those who were not able to do so, with a cut off score of 16 (sensitivity, 1.00; specificity, 0.89; classified, 0.99; positive likelihood ratio, 9.00) at t0 and a cut off score of 24 (sensitivity, 0.71; specificity, 0.69; classified, 0.70; positive likelihood ratio, 2.32) at one-year follow-up.

Ease of use and of maintenance of the device is commonly reported by patients and caregivers to be a significant factor in enhancing the AAC user’s experience [29, 31]. Although more sophisticated technologies like ETCD and BCI may become fundamental when, for example, eye movements are the only volitional mode left for communicating, the value of a sensor switch is its easy application because the set-up is simple and does not require any calibration procedure. Thus, to prolong AAC usage as much as possible in the course of the progressive neurodegeneration, we retain that a solution such as the LeMMS involving a single weak residual movement should be considered as preferential.

In contrast to commercially available sensors that are not specifically designed for PwALS, the LeMMS offers the advantage of requiring a selectable small activation force combined with an easy design to control communication with a minimum residual movement. The very low activation force required is a feature shared with capacitive sensors, but the LeMMS allows selection of the scanned item even in the absence of finger extension ability. This feature is unique to the LeMMS – it is not possible with capacitive sensors. Of course, this aspect is crucial in the presence of finger extensor problems.

Cooper et al. described, in AAC usage, issues with the battery running out, devices being broken or not working, or devices losing the proper set-up within a short span of time [32]. A low reliability creates discomfort both to the patient, who temporarily loses the ability to communicate, and to the caregiver, who needs to repeatedly set-up the system. Of course, the switch sensor is the most critical part susceptible to mis-positioning. In accordance with Spataro et al., the long daily usage of the AAC measured in our study (on average 5.45 h/day) is typical of regular users and testifies to the excellent reliability of the LeMMS [33]. In addition, no specific problems pertaining to reliability of the system were reported by caregivers during the follow-up period.

Family caregivers often describe difficulties with regard to the technical aspects of equipment requiring prompt intervention in the case of problems [31, 34]. This aspect is extremely important in the case of failure of the communication system and for this reason a secondary emergency solution is always available. In general, it consists of an emergency call bell and/or a transparent alphabet board. Obviously, the more sophisticated the AAC system, the greater the need for specialized support. In our study, no patient or caregiver made a specific request for support after the short training period. The long daily usage observed in our study combined with the LeMMS’s easy application should indicate that little support is needed from highly specialized and trained personnel. In fact, it is worth noting that, before the study commenced, all PwALS were already using an AAC system, so the only novelty in this study was the new type of sensor we provided them.

An important issue of the LeMMS is the easy positioning of the device for the caregivers. In fact, they need only to find the most appropriate finger position by moving the sensor, below it, until the proper leverage point, corresponding to the smallest activation force, is found.

It is well known that the involvement of the family in the palliative care process encourages family members to use AAC devices at home without this adding to their sense of burden [35]. Although in this study we did not address caregivers’ satisfaction via a specific questionnaire, the fact that no side effects or set-up/technical problems were reported during follow-up should imply that the caregivers had a positive perception of the new sensor. This positive perception could also be due to enhanced motivation as a result of their active involvement both in the training and follow-up.

The experienced therapists involved in this study during their intervention explored different technological solutions to allow our patients to continue to communicate. ETCD and BCI technology prescription were two possible alternative options to the LeMMS. It is well known that an access system such as eye control is faster than scanning-access for PwALS. In particular, Gibbons et al. [26] reported that PwALS were about three times faster with eye tracking than single switch scanning in writing a text.

However, the cost disparity between switch- and eye-controlled technologies is an important variable in clinical decision-making. Actually, the cost of high technology devices is considered a material barrier to their application and diffusion [14, 35]. In Italy, the cost ratio between scanning and eye control technologies is about 1 to 15, although this is expected to decrease thanks to recent technological advances [9]. Further, due to the frequent discrepancy between the focus of attention and the user’s direction of gaze (the “Midas touch problem”), some authors reported that error rates for PwALS were higher for eye control than for single switch scan [26, 36]. The same authors found no difference in the perceived efficiency between patients using a switch sensor or ETCD. In addition, there can be problems of accuracy with some ETCD devices and their complex calibration and set-up can complicate the handling [37]. Apart from patient-related factors (e.g. oculomotor dysfunction) that can impede ETCD application, the sensitivity of the infrared sensitive camera of ETCD technology to ambient infrared light reduces the usefulness of these devices in an outdoor setting. This points to LeMMS as the preferable choice overall for those patients who could benefit from a more portable version of the AAC device. In practice, we recommended ETCD only for those patients in whom the LeMMS did not work for communication.

Finally, BCI could in theory be an alternative to ETCD in PwALS in whom the usability of the AAC system is compromised by oculomotor dysfunction, gaze fatigue or loss of eye movement control. However, the feasibility of BCI for patients with such a rapidly progressing disease is undermined by the very long training sessions required to reach an adequate level of accuracy [14]. In addition, a long time is needed to set up an EEG-BCI [38]. For these reasons we did not consider BCI as a practical choice.