The participants in this study were identical to those who took part in the usability study of the mHealth app [34]. The data for the usability evaluation [34] and the current study (ie, swallowing training results) were collected from these participants in the same time frame. A total of 11 older adults (10 women; mean age 75.7 years, SD 4.96; mean education 10.4 years, SD 3.83; mean Korean Mini-Mental State Examination score 28.1, SD 1.72) from 2 district-run senior welfare centers in Seoul participated in the program. The inclusion criteria were (1) older than 65 years; (2) complaints of swallowing difficulties (ie, increased aspiration rate and foreign body sensation in throat); (3) a Korean Mini-Mental State Examination score within normal limits; (4) no problem with vision, hearing, and motor functions required for using a tablet; and (5) no hearing problems that may prevent the participant from following directions. The exclusion criteria were (1) any history of neurological disorders (ie, stroke and Parkinson disease) and (2) nonoral feeding (ie, nasogastric tube and percutaneous endoscopic gastrostomy). Self-reported swallowing rating was indicated as responses to the survey item “I have a swallowing problem” (with a scale of 0=never, 1=seldom, 2=sometimes, 3=often, and 4=always). Table 1 presents the characteristics of the participants. All participants gave informed, written consent as per the Declaration of Helsinki before conducting the study (institutional review board No. PIRB-2019-E024).

The app contents included a swallowing monitoring and intervention protocol with 3 therapeutic methods or maneuvers: (1) effortful prolonged swallowing, which incorporated effortful swallowing [35] with the Mendelsohn maneuver [19,36], (2) the effortful pitch glide exercise [37], and (3) effortful tongue rotation [29]. The 3 effortful maneuvers were named A Successful Swallow with Effortful Trainings (ASSET). Methods for each training maneuver used in this study are shown in Table 2.

Participants were required to perform a total of 120 sessions (3 times/day, 5 days/week for 8 weeks). A 1-day session involved 20 repetitions of each of the 3 exercises (effortful prolonged swallowing, effortful pitch glide, effortful tongue rotation) in the morning, afternoon, and evening, with each time-specific session taking approximately 30 minutes. If the individuals failed to conduct the exercises on time, they were allowed to make up for the missed sessions on the same day only.

We developed a tablet-based mHealth app (Android operating system) called 365 Healthy Swallowing Coach, which enables older adults to execute the given 3 types of swallowing training maneuvers without a clinician physically present. It contains an educational program, a feedback system, an adherence monitoring system, and a tailored training setting. The mobile app was designed to be downloaded and used on a tablet, the Samsung Galaxy Tab A (model No. SM-P580; Samsung), and to be user-friendly for older adults. Examples of user interface designs for older adults with decreased cognitive and psychomotor functions include (1) tabs arranged by use sequence for ease of navigation, (2) buttons containing both icons and text for ease of understanding, and (3) apposite button activation ranges, button sizes, and interbutton spaces empirically identified for the trade-off between ease of activation and prevention of inadvertent activation [34]. Furthermore, the mHealth app supports a variety of features that could enable older adults to properly and effectively train themselves. Examples of training compliance enablers include videos, animations, multimodal content, a biofeedback system for ease of understanding [38], and an automatic training data-logging system and training scheduler for ease of adherence monitoring [34]. A usability and feasibility test of the 365 Healthy Swallowing Coach app reported significantly higher usability scores from older adults with higher levels of education and smart device usage. Moreover, an increase in favorable responses regarding app acceptability, training program utilization, emotional responses, and learning experience with the intervention program was observed as the participants continued to use the app [34].

A flowchart of the 8-week home-based intervention study with biweekly (ie, every 2 weeks) human mediation is presented in Figure 6. Two weeks before the beginning of the intervention program, participants’ demographic information and medical histories were collected, and their isometric and swallowing tongue pressures were measured by a clinician. Subsequently, they were educated on how to operate the designated tablet and the swallowing function training contents of the 365 Healthy Swallowing Coach app before starting their home training using the mobile device (orientation stage). Upon commencing the intervention program, the researchers held biweekly face-to-face meetings with each of the participants for the purpose of checking the participant’s ability to carry out the training sessions without assistance and perform the exercises in accordance with the given protocol. In this way, the researchers acted as evaluators who checked and monitored the training status of the participants. However, since the primary purpose of the face-to-face meetings was strictly confined to monitoring the participants’ progress, the researchers minimized giving advice that might directly influence their self-administered performance. Each face-to-face meeting was held for approximately 30 minutes. During the meetings, the participants were asked to navigate through the app (ie, from the main screen of the tablet to the training selection screen) and perform 1 training session as an example while the researchers monitored their presentation. If the participants failed to operate the app properly or performed training inaccurately, they were advised by the researchers to try to correct themselves. After each participant finished a training session and closed the app, the researchers then checked the participant’s progress by reviewing the sessions tracked in their tablet.

At postintervention (ie, immediately after the 8-week intervention), the participants’ isometric and swallowing tongue pressures were evaluated, as previously performed during the preintervention evaluation stage. No additional intervention or intervention-related mediation took place after the postintervention evaluation stage. Then, at 12 weeks postintervention (ie, 20th week from the beginning of the intervention program), the participants’ isometric and swallowing tongue pressures were reevaluated in an identical manner as in the preintervention and postintervention evaluation stages in order to evaluate the maintenance effect of the intervention program.

Swallowing and isometric tongue pressures were measured using the Iowa Oral Performance Instrument (IOPI) (model 2.1; IOPI Medical LLC). The IOPI is a handheld device with a connecting tube and a measurement bulb (Figure 7) used to measure tongue pressure against the hard palate in an air-filled bulb.

The evaluations were conducted by the authors HK, N-BC, and SS. All participants had a verbal training session before having their perioral pressure recorded. During the tongue pressure measurement, the participants were in a comfortable and relaxed upright seated position to achieve a natural head posture, and the researchers oversaw their posture and the bulb position. When measuring the swallowing tongue pressure, the participants were asked to swallow the saliva as naturally as possible while the tongue bulb was placed between their tongue and the anterior one-third of their hard palate. When measuring the isometric tongue pressure, they were asked to push the bulb with maximum force toward their hard palate with their tongue for 2 seconds.

Each measurement was repeated 2 times. The pressure was recorded in kilopascals and the maximum value was retrieved. For swallowing tongue pressure, the participant was asked to perform a dry swallow [7]. Tongue pressure measurements were recorded before and after the intervention, as well as at 12 weeks postintervention. Immediate postintervention evaluation and 12-week postintervention evaluation findings were compared with preintervention evaluation findings to identify the short-term and long-term effects of the given therapy sessions. Statistical analysis was conducted using the Wilcoxon signed rank test (SPSS version 25; IBM Corp).

The adherence of all 11 participants was evaluated at the end of the 8-week intervention program. Table 3 presents the adherence of each participant. Out of the 11 participants, 8 completed the 8-week program with the proper amount of training suggested by the researchers (adherence of 83.2%-100.8%) and were categorized as compliant participants. Only these participants were included in the data analysis. Meanwhile, 3 participants (No. 9, No. 10, and No. 11) labeled as noncompliant participants were excluded from the analysis due to failure to correctly adhere to the program protocol. Participants No. 9 and No. 10 were excluded because they completed only 35.4% and 41.1% of the required sessions, respectively. The 2 participants reported that they did not have enough time to carry out the training sessions and that they could not perform each training session in accordance with the given protocol due to the excessive amount of training. Meanwhile, participant No. 11 was excluded due to greatly exceeding (264.7%) the amount of training required. She reported that she did not understand the adherence monitoring system of the app due to unfamiliarity with the technology, resulting in an excessive amount of training.

Table 3

Adherence of each participant.

The 8 participants who properly completed the 8-week training protocol, which was mediated by 3 sessions of biweekly meetings, demonstrated statistically significant increases in swallowing tongue pressure postintervention (median 26.5 kPa) compared with the preintervention (median 17.5 kPa) values (z=–1.992, P=.046). On the other hand, maximum isometric tongue pressures were not statistically different between the postintervention (median 41 kPa) and preintervention (median 40.5 kPa) evaluation (z=–0.632, P=.53). Table 4 and 5 show ranks and test statistics of swallowing and isometric tongue pressure before and after the intervention.

Of the 8 participants, 6 (No. 1, No. 2, No. 3, No. 5, No. 7, No. 8) agreed to participate in the 12-week postintervention evaluation for examination of maintenance effects. Compared with preintervention (median 19 kPa), there was no statistically significant difference in swallowing tongue pressure at the 12-week postintervention (median 20.5 kPa) (z=–0.734, P=.46). Table 6 presents ranks and test statistics of swallowing tongue pressure measured at the preintervention and 12-week postintervention evaluation stages.

To the best of our knowledge, this investigation is the first training study addressing an mHealth tongue-strengthening exercise for older adults. As tongue pressure measures have been determined to be effective indicators of swallowing function [7,8], an increase in tongue pressure may positively influence swallowing function as well. Decreases in swallowing tongue pressure generated during dry and high-viscosity liquid or bolus swallowing is prevalent in older adults [7,17]. Since such decline may be indicative of risk factors of dysphagia, looking at measures of swallowing tongue pressure in order to improve swallowing rehabilitation endeavors is a compelling pursuit [17]. In this study, a tablet-based swallowing training app with biweekly human mediation was used for 8 weeks by older adults who complained of swallowing difficulties. All 8 participants who completed the training demonstrated an increase in swallowing tongue pressure, even though therapists were not physically present during the training sessions. It should be noted that biweekly face-to-face meetings were held between the participants and the researchers for the main purpose of checking and monitoring the participants’ training progress, and this process may have affected the end results (ie, increased swallowing tongue pressure) of the intervention program. There are several important issues that must be recognized regarding these study findings.

Previous studies have shown that various tongue-strengthening exercises resulted in an increase in tongue strength and pressure [40-44]. Meanwhile, the current intervention contents (ASSET) adopted 3 therapeutic methods or maneuvers (ie, effortful prolonged swallowing, effortful pitch glide, and effortful tongue rotation) in combination, which led to positive effects on the swallowing tongue pressure of the participants, as indicated by the study results. However, several differences have been observed in treatment effects upon comparison of the current study to those that adopted a single therapy maneuver.

First, the current study implemented effortful prolonged swallowing, a combination of effortful swallowing training and the Mendelsohn maneuver training. In previous studies, training outcomes of effortful swallowing alone only revealed significant increases in the maximum isometric pressure of the anterior tongue, lingua-palatal pressures, and swallowing functions [41,45]. However, in this study, effortful prolonged swallowing positively affected swallowing tongue pressure. The functions of the extrinsic tongue muscles (ie, hyoglossus), which were shown to be active during the Mendelsohn maneuver [46], may have improved after therapy, thus benefiting the participants and positively affecting their swallowing tongue pressure.

Second, effortful tongue rotation is regarded as a promising training method, since swallowing tongue pressure is closely related to dysphagia symptoms such as aspiration [9] and oropharyngeal residue [47]. A previously conducted study observed a significant increase in maximum isometric tongue pressure after effortful tongue rotation [29]. Interestingly, however, only the swallowing tongue pressure and not the maximum isometric tongue pressure was enhanced in this study when the isokinetic effortful tongue rotation exercise was implemented.

Partial discrepancies observed among the previous and current results may be ascribed to the fact that the triple-combination therapy (ASSET) of effortful prolonged swallowing, effortful pitch glide, and effortful tongue rotation was employed in this study. Hence, the combination effect is not yet comparable to that of a single therapy maneuver. Consequently, the results from this study protocol are not indicative of the individual effect of each treatment method on posttreatment measures, since the treatment effects of each of these methods were not evaluated separately. Therefore, it cannot be concluded that the additive effect of the triple-combination therapy as a whole is equal or similar to the arithmetic addition of each individual exercise’s training effect. The effect that the 3 combined methods yield may be smaller than the effect that each method provides, or it may be greater [48] than the simple addition of individual effects.

Even though a significant effect on swallowing tongue pressure was obtained from this intervention program, it remains unclear if the actual cumulative exercise dosage (ie, intensity, frequency, and duration) employed in this study is significant [49]. According to a scoping review, a tongue exercise dosage of 3 times a day for 7 days with a frequency of 30 repetitions for each exercise was the most widely implemented regimen for studies including single tongue exercises [50]. However, for studies that adopted combined exercises (ie, more than one therapy method), treatment dosage, frequency, and duration varied widely, and thus, the improvements in various outcome measures could not be concluded to be attributable to the use of particular treatment dosages [50].

Determining the optimal exercise dosage of a treatment program is crucial, as overtraining may contribute to exercise-induced fatigue [51] and high dropout rates [52], whereas undertraining may not result in any benefit [53]. In fact, some participants in this study complained of the excessive training load and stated that they could not complete the intervention program.

The “active ingredients” of interventions need to be identified in order to determine the optimal intervention dosage and intensity [49]. The active ingredients are components in an intervention that can influence the treatment results. Client-related variables, including life situations or circumstances, or therapy-related variables, such as the complexity of treatments and the treatment duration [54], may also influence the optimal therapeutic dosages. Since the dosage stated in repetitions (eg, 20 exercise repetitions per session) indicates how many times an “active ingredient” appears per training session, it may be important to represent exercise dosage in such a form when determining the optimal intervention dosage for speech language pathology–related fields [49].

Choosing to adopt a distributed intervention or an intensive intervention is also critical in determining the ideal intervention dosage and intensity. Previous studies have adopted intervention programs for various speech and language disorders with different levels of intensity [55-57]. One study that compared the intervention results of a distributed (8 weeks, 6 hours per week, for a total of 48 hours) and intensive (3 weeks, 16 hours per week, for a total of 48 hours) aphasia intervention showed that the distributed intervention derived greater therapeutic gains compared with the intensive treatment [55]. The protocol duration implemented by this study is comparable to the distributed intervention adopted in the aforementioned study because the total intervention duration of this mHealth program was also 8 weeks (20 repetitions of each exercise, totaling 60 repetitions/session, 3 sessions/day, 5 times/week). However, the exact dose per session, stated as the number of repetitions, was not indicated in the aphasia intervention study. Thus, an exact one-to-one comparison is not possible.

The current study measured the swallowing tongue pressure at baseline and at 2 follow-up time points, including at 20 weeks after the intervention commenced (ie, 12-week postintervention), in order to investigate long-term intervention effects. Tracking changes in gained muscle (ie, tongue) pressure after the removal of an intervention is of particular interest, since a consideration of detraining effects may be necessary for the effective planning of aftercare [30]. In the current study, the training effect in the swallowing tongue pressure observed at the termination of the 8-week intervention program did not remain upon measuring it at 12 weeks postintervention. Despite the clinical implications of detraining effects, however, studies that examine the long-term effects of swallowing interventions are largely lacking [30,58]. Past studies show inconsistent results in posttreatment decay of lingual strength. Discrepancies in the findings from previous studies may be ascribed to several methodological differences, including (1) the postintervention measurement points, (2) the age of the participants, (3) the baseline health conditions of the participants, (4) the intervention delivery method, and (5) the type of outcome measure used.

The first factor is the postintervention measurement points. When the current results are compared with those of past studies, the long-term efficacy of tongue-strengthening exercises does not seem to show a consistent correlation with the measurement point alone. For instance, significant decreases in tongue strength have been observed after 2 to 4 weeks [30] and after 12 weeks (in the current study) of detraining, while no changes or decreases have been found at 4 weeks [59] and 28 weeks [40] postintervention. Since participant details and intervention dosages varied across the studies, the sole effect of postintervention measurement points needs to be addressed in a separate controlled study.

The second and third factors are the age and the baseline health conditions of the participants. A common observation is that older adults are more vulnerable to posttreatment decay [60,61]. Moreover, detraining effects are known to be larger for those who exhibit more impaired swallowing than those with more functional swallowing at baseline [58]. Hence, we hypothesize that the relatively high age and the swallowing difficulties of this study’s participants may have contributed to their markedly decreased posttreatment performance compared with the aforementioned studies, which targeted healthy adults without swallowing problems.

The fourth factor is the intervention delivery method. This study employed a tablet-based mHealth app for a swallowing intervention, which fundamentally differs from traditional methods, as telepractice requires participants’ self-management with the technology and their active engagement [62] for higher performance. Hence, it is possible that the long-term effects of telepractice interventions are susceptible to individual-dependent aspects (eg, technology learning rate, engagement rate).

The fifth factor is the type of outcome measure used (ie, isometric tongue pressure vs swallowing tongue pressure). As seen from the results, the comparison between isometric and swallowing tongue pressure demonstrated that only the latter was significantly improved in the posttreatment performance. In addition, both tongue-related measures returned to baseline after the detraining period, providing no direct clues regarding the correlation between a specific type of measure and its long-term effect. Although the maintenance of different types of measures is rarely found in the previous literature, the discrepancy between the posttreatment performance of isometric versus swallowing tongue pressure in this study indirectly suggests that the chosen outcome measure may be one possible factor that needs to be considered when interpreting long-term postintervention effects.

The mHealth platform can enhance the delivery of exercise maneuvers through improvements in training accuracy, economic efficiency, safety continuity, and self-management compared with traditional clinician-to-client intervention [63-66]. In this study, we used a biofeedback system involving a mirror function (in effortful prolonged swallowing and effortful tongue rotation) and a real-time pitch graph (in effortful pitch glide), preventing improper or incorrect performance, which may counteract the purpose of the maneuvers and become entrenched [67,68]. In this way, the training accuracy is maintained. Although there is no sufficient evidence that biofeedback benefits overall swallowing functions among patients with dysphagia, in studies that used tongue manometry as biofeedback, significant increases in maximum isometric and swallowing tongue pressure were observed in patients who received biofeedback compared with those who did not [69]. In addition, a previous study has suggested that home-based therapy can be as effective as office-based therapy if a biofeedback system is applied [70]. In addition to increasing training accuracy, the home-based mHealth training app may enable the participants to attend various training programs (eg, the 8-week program adopted in this study) without the burden of visiting clinics as frequently or the need to pay extra costs [71,72]. Thus, the time- and cost-effectiveness of mHealth programs with tongue-strengthening exercises may represent strong points for promoting their widespread use. Furthermore, older adults, including community-dwelling and patient groups, have higher risks for falls and disease complications, which highlights the need to promote their safety by reducing the frequency of their clinic visits [65,73]. The mHealth service could also allow for continuation of therapy sessions even during unexpected circumstances, such as pandemics (eg, COVID-19) [74]. A nondirect clinician-to-client intervention service could serve as a valid alternative to direct interaction.

In addition, it is highly necessary to construct mHealth tools that provide the support needed to promote participants’ adherence to the program. In general, the training adherence rate among the older population with various chronic diseases tends to decrease significantly for long-term training [75]. To address this, the tablet-based app used in this study included reminder functions, such as a calendar, a progress bar, and alarms, to promote participants’ self-monitoring of their adherence level. For instance, if a participant did not meet the training goal in a session, an incomplete red (or orange) circle would appear on the calendar and the progress bar would not be filled completely in green. The participant could then check their progress through visual biofeedback and finish the training accordingly. Such reminder features are commonly adopted in other training apps [76,77] and may be considered a crucial component of a self-therapy–related app. App features aimed at promoting users' adherence (eg, reminders, feedback), however, may not be designed to help users easily maintain long-term engagement without external support [78-80]. This is especially true for older adults who are not familiar with digital devices and apps, including mHealth devices [34,81]. Therefore, a hybrid method of intervention delivery (eg, mHealth app intervention with human mediation) may be necessary to help older adults maintain their adherence and thus potentially maximize the intended therapeutic effect, even when using the mHealth app as a primary means of swallowing health care. In this study, researchers checked the participants’ progress face-to-face once every 2 weeks to monitor their adherence to the intervention program using the mHealth app. A total of 8 of 11 participants successfully completed the program with optimal intended adherence, and it is possible that the overall adherence of the participants may have been lower in the absence of human mediation (ie, without the biweekly face-to-face meeting). A further study is warranted to see if similar therapeutic results would be observed for the mHealth intervention without external support.

Furthermore, in order to compensate for the possible issues caused by face-to-face interactions and to develop the current app platform into a more convenient means of intervention, it is suggested that real-time monitoring and control solutions be added into the app to allow clinicians to remotely keep track of the participants’ adherence and performance.

The findings from the study indicate that, using the mHealth app, the intervention was able to help older adults increase their swallowing tongue pressures. However, this study is not without limitations. This study involves a small number of participants, which does not permit the generalization of its findings. Accordingly, we could not classify participants by severity and make group comparisons. Furthermore, only 6 participants agreed to take part in the 12-week postintervention evaluation. The results from these participants may not have been sufficient to derive a generalized long-term effect of the given protocol. Thus, an implementation of the mHealth app with a larger sample size is proposed. Finally, although the effect of biweekly face-to-face monitoring on the increased swallowing tongue pressure measured at the postintervention phase is not negligible, the current study did not address the independent effect of the face-to-face monitoring on mHealth intervention effects.

We suggest additional issues that may be useful to address in future studies. First, although the participants of the current study did not receive any retraining during the 8 weeks of rest, it is likely that subsequent retraining may help participants retain their swallowing tongue pressures. Hence, a randomized controlled trial is further warranted to see if a treatment group (which follows a maintenance protocol after the intervention) exhibits higher long-term effects of training compared with a control group (which does not receive any retraining after the intervention). In addition, it may be necessary to take into consideration the frequency of postintervention measurements and the measurement points (eg, 8 weeks versus 12 weeks postintervention) to see if the detraining period influences the maintenance of swallowing functions.

Furthermore, the training intensity may have negatively affected the participants’ overall adherence. Several participants in our study expressed dissatisfaction toward the training intensity and frequency, claiming that 3 sessions a day is excessive, given that they have other plans during the day. Although the 365 Healthy Swallowing Coach app originally provided a tailored training setting for users (ie, users are able to customize the training schedule and training amount), this feature was made unavailable in the current study to measure the therapeutic effect of the fixed amount of training (ie, 3 times/day, 5 days/week). Hence, the reported problems related to the intensity of the protocol call for additional research to systematically investigate the optimal dosage of the mHealth swallowing intervention by comparing the effect of intensive (eg, 3 times/day, 7 days/week for 2 weeks) versus distributed (eg, 1 time/day, 3 days/week for 12 weeks) intervention methods.

Swallowing tongue pressure is known to be closely related to dysphagia symptoms. This is the first study to demonstrate the effectiveness of the combined methods of effortful prolonged swallowing, effortful pitch glide, and effortful tongue rotation using mobile app training accompanied by biweekly human mediation in improving swallowing tongue pressure in older adults. The fixed amount of the training dosage (ie, 3 times/day, 5 days/week) was monitored biweekly to maintain the participants’ optimal adherence to the training program. As a result, a significant increase in swallowing tongue pressure was observed after 8 weeks of the intervention. This mHealth pilot study will help guide future intervention studies for older adults who have limited access to health care services. The mHealth app program is a promising method of providing therapy through the execution of intensive and repetitive tongue exercises for vulnerable older adults. In order to observe therapy efficacy with a larger sample size and to investigate long-term effects of the intervention program, further studies are warranted. Moreover, further studies using videofluoroscopic swallowing studies are needed to investigate the effects that the app-based therapy program may have on the various stages of swallowing.