Feasibility and Physiological Effects of a Home-Based Swallow Exercise Program Using sEMG Biofeedback in Prefrail Older Adults: A Case Series

Ali Barikroo ^1,* , Alyssa Zinser ²

1. Swallowing Physiology & Rehabilitation Research Laboratory, Speech Pathology and Audiology Program, Kent State University, Kent, Ohio, USA

2. Craig Hospital, Englewood, Colorado, USA

* Correspondence: Ali Barikroo 

Academic Editor: Arthi Veerasamy

Special Issue: Oral Hypofunction and Oral Frailty in Older Adults

Received: April 28, 2025 | Accepted: August 03, 2025 | Published: August 13, 2025

OBM Geriatrics 2025, Volume 9, Issue 3, doi:10.21926/obm.geriatr.2503322

Recommended citation: Barikroo A, Zinser A. Feasibility and Physiological Effects of a Home-Based Swallow Exercise Program Using sEMG Biofeedback in Prefrail Older Adults: A Case Series. OBM Geriatrics 2025; 9(3): 322; doi:10.21926/obm.geriatr.2503322.

© 2025 by the authors. This is an open access article distributed under the conditions of the Creative Commons by Attribution License, which permits unrestricted use, distribution, and reproduction in any medium or format, provided the original work is correctly cited.

1. Introduction

Age-related dysphagia affects a significant proportion of aging individuals, with a mean prevalence of 15% reported in community-dwelling older adults [1]. Sarcopenia—characterized by the progressive loss of muscle mass and strength—is a leading contributor to age-related dysphagia [2], which can lead to malnutrition, aspiration pneumonia, and even death [3]. Prophylactic strength-based swallowing exercises can potentially be used as part of a multi-prong approach to mitigate the risk of dysphagia in older adults. Existing evidence suggests that strength-based swallowing exercises, such as the effortful swallow (ES) maneuver, can improve different kinematic and physiological aspects of swallowing, including maximum lingual-palatal pressure, maximum pharyngoesophageal segment (PES) opening duration, superior and anterior hyolaryngeal excursion, posterior pharyngeal wall thickness, and an increase in pharyngeal contractile integral [4,5,6,7,8].

Despite these positive benefits, access to these exercises for aging individuals is limited because of the intensive nature of the regimen, requiring several weeks and a 5-day-a-week commute to a clinic. For many older adults—particularly those with physical limitations or living in underserved areas—these requirements are impractical due to transportation and mobility barriers [9]. While home-based delivery could increase access, its benefits may be constrained by the absence of sophisticated biofeedback tools needed to guide performance [10]. Moreover, two systematic reviews have highlighted that home-based exercise programs are particularly vulnerable to non-adherence due to the absence of supervision and the reliance on potentially inaccurate self-reported exercise logs [11,12].

Home-based digital health tools that incorporate real-time biofeedback offer a promising alternative to improve accessibility and adherence to swallowing exercises [13]. Surface electromyography (sEMG) biofeedback, in particular, has been widely used to enhance swallowing therapy by providing users with real-time visual feedback on muscle activity. This approach facilitates training by enabling individuals to monitor and adjust their effort during exercises such as the ES maneuver [14]. The ES maneuver is commonly used in dysphagia rehabilitation to target the lingual, floor-of-mouth, and pharyngeal muscles—key components of the swallowing mechanism. As such, incorporating this maneuver into an exercise regimen may help mitigate sarcopenia by improving muscle strength and motor unit recruitment, thereby supporting preservation of swallowing function [7].

To address barriers related to accessibility, adherence, and the lack of biofeedback, this study integrated sEMG biofeedback into a home-based swallowing therapy program using the Mobili-T^® system. The Mobili-T^® device is a compact, wireless sEMG system that attaches to the submental area and connects via Bluetooth to a smartphone application. This app provides real-time visual biofeedback on the intensity of submental muscle contraction during each ES exercise and displays success rates for individual trials. Participant performance data are automatically recorded and securely transmitted to a dashboard accessible by the assigned SLP clinician for monitoring adherence and support. Prior studies have demonstrated the feasibility and usability of the Mobili-T^® system in populations such as head and neck cancer survivors [15] and stroke patients [16], showing improved adherence and functional outcomes. In addition to usability, the system has demonstrated sound psychometric properties. Its automated swallow-detection algorithm achieved high sensitivity (92.3%) and positive predictive value (83.9%) in healthy adults. Test–retest reliability was also acceptable, with ICC value of 0.606 [17]. These findings support the validity and reproducibility of Mobili-T^® for capturing submental muscle activity in remote swallowing interventions. However, limited evidence exists on whether this technology can be incorporated as a home-based prophylactic swallow exercise paradigm for older adults.

This case series served as a first step to identifying the primary efficacy and feasibility of such a program. Thus, we aimed to evaluate the effect of home-based ES training using this biofeedback technology on lingual strength and submental muscle activity along with different aspects of feasibility including adherence and success rates as well as participant satisfaction in three prefrail older adults without dysphagia.

2. Materials and Methods

2.1 Study Design

This pretest-posttest case series investigated the preliminary efficacy (Aim #1) and feasibility (Aim #2) of a home-based prophylactic swallow exercise program using the Mobili-T^® system in three prefrail older adults.

2.2 Study Procedure

2.2.1 Screening Measures

Three potential participants underwent a comprehensive screening procedure to ensure eligibility criteria. Their medical history was first obtained through a medical history questionnaire, which served to identify any pre-existing conditions that could impact swallowing function. In addition, several standardized measures were utilized to assess demographic and physical health factors. The Eating Assessment Tool-10 (EAT-10) [18], a validated self-administered dysphagia screening questionnaire, was used to evaluate self-reported swallowing difficulties. Potential participants were expected to have an EAT-10 score of 2 or less, as scores above this threshold are indicative of potential dysphagia. To assess overall physical status, the Fried Frailty Index [19] was administered, incorporating self-reported measures of unintentional weight loss, exhaustion, slowness, activity level, and weakness. Potential participants were required to score 1 or 2 out of 5 to be classified as pre-frail and to meet the inclusion criteria to participate in this study. Cognitive status was screened using the Modified Mini-Mental State Test (3MS), a standardized tool for evaluating cognitive function [20] (Table 1). In addition, the potential participants completed the 3-oz Water Swallow Screening Test [21], in which they drank 90 ml of water uninterrupted while being monitored for any signs of swallowing difficulties. These measures collectively ensured that only appropriate candidates were included in the study. Lastly, submental adipose tissue thickness (SATT) was measured to explore its potential influence on sEMG signal detection. Although no participants were excluded based on this factor, it was included as an exploratory variable to better understand how SATT may impact sEMG signal quality. Details of the measurement procedure have been described in our previous publications [22,23]. All three individuals successfully met the eligibility criteria and were enrolled in the study.

Table 1 Participant characteristics including demographics, cognitive function, frailty status, and swallowing screening.

2.2.2 Procedure

The study was conducted in three phases: baseline measurement and instruction, intervention phase and tracking of adherence, and post-intervention assessment and trial reflection.

Baseline Measurement and Instruction Phase. Following the collection of baseline physiologic measures, participants received instruction on performing home-based swallow exercises. The investigator demonstrated the installation and navigation of the Mobili-T app on their smartphones. The participants were also trained in setting up the device, positioning it under their chins, and ensuring proper adhesion (Figure 1). To confirm their ability to use the device independently, they were asked to operate it without assistance. This step served as a functional screening to ensure participants could smoothly use the smartphone app and device without external support. Participants who were unable to independently set up and navigate the app would have been excluded from the study, although this was not necessary as all screened participants successfully completed this step.

Figure 1 Mobili-T^® system set up and placement.

Intervention Phase and Tracking of Adherence. The intervention phase lasted six weeks, with participants performing ES exercises at home five days per week using the Mobili-T device. This device provided real-time feedback, displaying successful and unsuccessful effortful swallows as dark and light-colored cubes on the screen. The participants could view waveforms and a target bar to adjust their swallowing efforts accordingly. The Mobili-T program automatically counted the number of swallowing attempts and recorded both successful and unsuccessful swallows in real time, eliminating the need for participants to manually track their exercise performance. To minimize the potential influence of non-swallow-related muscle activity (e.g., tongue or jaw movements) on the accuracy of the automated swallow count, participants were explicitly instructed to keep their head still and in a neutral, upright position during the exercises and to avoid additional movements, including tongue and jaw motions not involved in the swallow. These instructions were emphasized during the initial training and reinforced throughout the intervention period to ensure consistency and optimize measurement reliability. Exercise data were securely transmitted to a secure dashboard, allowing the research team to monitor adherence and performance.

Each exercise session began with an automated calibration process, where the participants sat still and performed three regular dry swallows. The software then set the success threshold for effortful swallows at 20% above the average sEMG peak recorded from these swallows. During each session, the participants completed eight sets of nine repetitions of dry effortful swallows, totaling 72 trials. A 10-second rest period was automatically programmed within the Mobili-T app between trials, and participants were trained to manually time a two-minute rest interval between sets using a separate timer. Each session lasted approximately 30 minutes. To prevent dry mouth, participants were permitted to take sips of water between trials if needed. Weekly video calls were conducted to verify proper device setup and correct execution of exercises. Troubleshooting assistance was provided as needed.

Post-intervention Assessment and Trial Reflection. After completing the 6-week intervention, participants returned for an in-person session, during which all baseline physiologic measures were reassessed. Following this assessment, participants completed a semi-structured interview to provide detailed insights into their experiences and perceptions of the program. A consistent set of eight open-ended questions was asked of all participants to ensure comparability across responses while allowing for elaboration and individual nuance (See Table 2). Each interview lasted approximately 4-6 minutes, was audio recorded, and transcribed verbatim by two investigators subsequently for later qualitative analysis.

Table 2 Post-Intervention Interview Questions.

2.3 Outcome Measures

2.3.1 Physiologic Measures

Electrical Activity of Submental Muscles. sEMG was used to record changes in the electrical activity of submental muscles during tongue press, normal swallows, and effortful swallows. The sEMG signals were recorded using the ACP Synchrony^® system (Accelerated Care Plus Corporation, Reno, NV). The reliability and validity of sEMG for identifying swallowing events from the submental region have been supported by multiple studies. Crary et al. demonstrated high classification accuracy (90%) and excellent inter-rater reliability (kappa = 0.80) for trained raters distinguishing swallows from nonswallow movements based on sEMG signals [24]. Furthermore, biomechanical validation work by Crary et al. [25] showed a strong temporal correspondence between submental sEMG activity and key biomechanical swallowing events—particularly hyoid elevation and pharyngoesophageal segment opening—establishing the physiological relevance of sEMG signals during swallowing. Poorjavad et al. [26] also reported moderate to high test–retest reliability for submental sEMG amplitude during swallowing tasks, supporting the reproducibility of this measure across sessions. A series of strategies to mitigate the inherent variability associated with sEMG were implemented. Standardized electrode placement, consistent participant posture, rigorous skin preparation, and control of the measurement environment were applied uniformly across all sessions to reduce potential variability.

A rectangular electrode disk measuring 29 mm × 87 mm, with two embedded electrodes of 10 mm in diameter, was used for data collection. Before data collection, the participants’ submental region was cleansed with an alcohol pad, and the sEMG electrode disk was placed at the center of the submental area, between the mandible and hyoid bone. To ensure accurate signal capture, participants performed a dry swallow prior to recording. sEMG data were collected while they completed three maximum lingual-palatal press tasks with IOPI at the anterior and posterior tongue followed by three swallows each of normal and effortful swallows across varying 10 ml bolus consistencies including thin liquid (water), pudding-thick liquid, and solid. Participants were instructed to "swallow the material in one swallow if possible."

For effortful swallow, participants were instructed to push their tongue against the roof of their mouth and swallow with maximum effort, simulating the sensation of swallowing a large piece of steak. All the materials (except solid) were presented via a 10 ml syringe. Before initiating a swallow, they were instructed to notify the investigator using a hand signal. The investigator started and paused the sEMG recording at the beginning and end of each swallow trial, respectively. The recorded sEMG traces were then labeled and saved for offline analysis.

To extract the sEMG data, the labeled sEMG waves were identified first, and the onset and offset points of each wave were visually marked. The onset point was defined as the moment a sharp increase in sEMG amplitude was observed, while the offset point was defined as the moment the trace returned to its resting level. The sEMG average amplitudes for each swallow were automatically extracted using the incorporated software in the ACP Synchrony^® system in microvolts (µV). All sEMG and tongue pressure measurements were performed by the same trained examiner following standardized protocols to minimize variability. The examiner also reviewed all sEMG traces offline to ensure consistent identification of onset and offset points, supporting measurement reliability throughout the study.

Tongue Strength. Tongue strength was measured using the Iowa Oral Performance Instrument (IOPI, model 2.1; IOPI Medical LLC, Carnation, WA, USA). Participants were instructed to press their tongues against an air-filled bulb, which was positioned against the roof of their mouths. This bulb was attached to a pressure recorder that provided a measure of lingual-palatal pressure in kilopascals (kPa). Three maximum pressure recordings were obtained from both the anterior and posterior tongue, with the highest value recorded as the maximum pressure. Additionally, simultaneous sEMG recordings were taken during anterior and posterior lingual-palatal pressure measurements to capture the associated submental muscle activity.

2.3.2 Feasibility Measures

Adherence. The embedded software calculated the percentage of attempted effortful swallows compared to prescribed doses and generated daily, weekly, and overall adherence percentages.

Success Rate. The software reported the success rate of the intervention by dividing the number of successful effortful swallows by the total number of attempted swallows.

Satisfaction. At the end of the exercise period, semi-structured qualitative interviews were conducted to explore their experience with the program. The interviews were recorded, transcribed, and analyzed.

2.4 Statistical Analysis

The minimal detectable change (MDC) thresholds used in this study were adopted from previously published reliability studies. Specifically, the 17.5% and 16% thresholds for anterior and posterior tongue pressure, respectively, were derived from Adams et al. [27], who established these values based on repeated testing and reliability analyses in healthy adults. The 50% threshold for submental sEMG amplitude was based on Poorjavad et al. [26], who reported this percentage as a conservative marker for meaningful physiological change in surface electromyography swallowing assessments. These thresholds were used descriptively to determine whether observed changes exceeded the expected range of measurement error. Adherence and success rates were automatically calculated by the device’s embedded software. Descriptive statistics and graphics were used to report the adherence rate to the exercise program, and participants’ success rates were reported as weekly and overall percentages. Participants were also interviewed post-intervention to assess perceived impact and usability. A qualitative descriptive approach was used, and transcripts were analyzed using an inductive, data-driven thematic analysis to identify patterns related to satisfaction, usability, and barriers to adherence. Two researchers independently reviewed and coded the transcripts, compared codes, and refined categories through discussion. Final themes were developed by consensus. While data saturation was not expected due to the small, predefined sample size (n = 3), the goal was to capture diverse participant perspectives to inform future refinement of the intervention.

2.5 Ethics Statement

The study protocol was approved by the local Institutional Review Board (approval number: 505) on February 17, 2023, and all the participants signed an informed consent form.

3. Results

Subject 1 demonstrated an adherence rate of 99.2% with a success rate of 68.3%. Her anterior (29.4%) and posterior (87.5%) tongue pressure gains exceeded MDC, indicating true strength improvements. However, none of her submental sEMG amplitude changes across multiple consistencies during regular swallows (6–25%) exceeded MDC (Figure 2). While she did not perceive a functional change, she noted greater awareness of her swallowing behaviors and reported the device was easy to use after the initial learning phase.

Figure 2 (a) Changes in the mean maximum lingual-palatal pressure (left) and submental sEMG amplitude average across different consistencies during regular swallows (right), from baseline to post intervention, (b) Changes in the mean synchronized submental sEMG amplitude average during anterior and posterior tongue press tasks (left) and submental sEMG amplitude average across different consistencies during effortful swallows (right), from baseline to post intervention (data available for Subject 3 only) and (c) weekly adherence and success rates for each participant throughout the intervention phase. The asterisk (*) indicates significant differences exceeding the MDC threshold.

Subject 2 achieved an adherence rate of 99.2% with a success rate of 68.3%. Her tongue pressure changes (12.5% anterior, 6.5% posterior) did not meet MDC thresholds, suggesting no measurable improvement in strength. However, submental sEMG amplitude during dry, thin, and solid swallows exceeded 50%, indicating true neuromuscular adaptation. She had initial technical issues, which were resolved, and appreciated the routine structure (Figure 2). Despite early technical issues with app connectivity and electrode placement, these were resolved with support. She did not perceive swallowing improvements but appreciated the program’s structure and routine.

Subject 3 exhibited an adherence rate of 94.3% of sessions and had the lowest success rate (48.2%) among participants. Her anterior tongue pressure increased by 19.5% (exceeding MDC), while posterior pressure rose by only 3.7% (below MDC). Submental sEMG amplitude increased >50% during dry, thin, and solid swallows, suggesting real adaptation, though pudding swallow changes (~29%) remained below MDC. She was the only participant with analyzable synchronized sEMG during tongue press and effortful swallows. There was a system error that led to missing synchronized sEMG recordings during tongue press and effortful swallows for Subjects 1 and 2. This issue was identified and resolved prior to the assessment of Subject 3, which allowed for complete data capture in this case. sEMG amplitude during the tongue press increased by 59.3% (anterior) and 105% (posterior), exceeding MDC. However, effortful swallows themselves showed only 10–41% increases—below MDC (Figure 2). She reported no functional changes but remained highly engaged and motivated.

4. Discussion

This case series provides early evidence supporting the feasibility and physiological promise of a home-based, biofeedback-supported ES training program for prefrail older adults. Despite some variability in individual outcomes, all participants completed the program with high adherence and reported it was usable and acceptable, reinforcing the potential of telehealth-administered dysphagia interventions.

To show that the reported changes exceeded the measurement variability, we linked observed changes in outcome measures to established clinical benchmarks. Specifically, we used published MDC thresholds to evaluate whether improvements in tongue pressure and submental sEMG amplitude exceeded expected variability. For example, Subjects 1 and 3 showed tongue pressure increases that exceeded the MDC thresholds of 17.5% (anterior) and 16% (posterior), with gains ranging from 19.5% to 87.5%. These values exceeded the respective benchmarks by 2–71.5%, indicating changes beyond expected measurement variability [27]. Additionally, Subjects 2 and 3 demonstrated increases in submental sEMG amplitude that exceeded the 50% MDC threshold across multiple consistencies during regular swallows, as well as during tongue press in Subject 3 (the only participant for whom this measure was collected). Gains ranged from 50% to 105%, indicating physiologically meaningful neuromuscular adaptation beyond expected variability [26]. Our findings align with existing literature demonstrating the neuromuscular benefits of ES when incorporated into structured training programs. The inclusion of real-time sEMG biofeedback likely enhanced user engagement and motor learning, echoing prior studies that have emphasized the importance of feedback in optimizing swallowing exercise performance [4,28,29]. By leveraging the Mobili-T^® device, our program enabled participants to visualize muscle activation and receive immediate performance-based cues—factors that may contribute to improved adherence and training efficacy, particularly in unsupervised home settings.

Notably, our remote protocol achieved physiological improvements comparable to in-person interventions reported in the literature [5,30]. These results suggest that a targeted, feedback-supported ES protocol can be delivered effectively in a home-based format without compromising training intensity or user engagement. This is further supported by parallels with Kim et al.’s [31] mHealth intervention, which also reported improvements in swallowing physiology using a tablet-based app and biweekly coaching. Although our protocol was less complex and involved no direct therapist mediation, similar trends in muscle adaptation were observed.

The lack of perceived functional improvement among participants is not surprising, as the intervention was designed as a prophylactic strategy rather than a rehabilitative one. All participants were asymptomatic at baseline, and the goal was to increase muscular reserve rather than address active swallowing impairments. This distinction is important: swallowing is a submaximal task, and improvements in maximum strength may not manifest in noticeable changes during everyday eating. Nevertheless, strengthening the muscles involved in swallowing may help delay or prevent future onset of sarcopenic dysphagia, particularly as age-related declines continue.

Additionally, our findings highlight the importance of individualized intervention dosing. Although the device recalibrated at each session by setting the success threshold at 20% above the average peak amplitude of regular swallows, which provided some degree of individualization, this fixed-threshold biofeedback algorithm may have been insufficiently challenging for individuals with higher baseline activity or those who demonstrated early neuromuscular gains. While the session-specific recalibration accounted for day-to-day variability, it did not incorporate a progressive increase that would adequately reflect cumulative improvements over time. Adjusting biofeedback thresholds using dynamic or progressive algorithms that adapt based on individual performance trajectories or clinical calibration may further enhance neuromuscular responsiveness and training efficacy. Moreover, longer intervention durations or optimized dosing regimens may be required to induce meaningful changes, especially in individuals with greater initial strength or functional reserve. Another noteworthy observation was the occasional disconnect between gains in tongue pressure and submental sEMG amplitude. While both are markers of swallowing-related muscular adaptation, their divergence in some cases suggests they may reflect different physiological processes or recruitment patterns. Submental sEMG primarily captures activity from the mylohyoid, geniohyoid, and anterior belly of the digastric muscles, while tongue pressure tasks mainly engage the genioglossus muscle, with limited submental muscle involvement [8,32]. Future research should explore the relationship between these measures to better inform targeted training strategies and more precisely guide dysphagia rehabilitation protocols. An important consideration for future research is whether long-term adherence to prophylactic swallowing exercise programs can be sustained when participants do not perceive immediate or noticeable functional improvements. Understanding the motivational factors that drive continued engagement in the absence of perceived benefit will be essential for designing effective, sustainable interventions aimed at preventing dysphagia in aging populations.

Consistent with a growing body of mHealth research, technical issues were minimal and manageable. Participants’ willingness to persist through occasional connectivity problems and continue training speaks to the acceptability of the approach. This aligns with findings from Hwang et al.’s [33] scoping review, which emphasized the value of feedback, reminders, and ease of use in sustaining older adults’ engagement in digital health interventions. Subject 3 exhibited a notably lower success rate compared to the other participants, which warrants further examination. A plausible contributing factor is the greater submental subcutaneous fat tissue thickness observed in this participant. Increased adipose tissue in the submental region may have attenuated the sEMG signal and diminished the sensitivity of the device in accurately detecting swallowing-related muscle activity. The interposition of soft tissue between the muscle and the electrode can weaken signal transmission, potentially accounting for the reduced success rate observed, despite the participant’s high adherence to the prescribed exercise protocol.

While the intervention demonstrated feasibility and potential benefits, considerations for broader dissemination are warranted. The cost of the Mobili-T^® device at the time of study was approximately $700 per unit, which is typically covered by SLP providers. Specifically, patients can access devices through established loaner programs that facilitate home-based practice without requiring direct purchase. Currently, insurance covers rehabilitation costs in North America but does not consistently cover preventative swallowing exercises. Future studies should focus on providing additional evidence for the efficacy of preventative approaches to support insurance coverage expansion and promote early intervention to mitigate dysphagia risk.

4.1 Limitations

Interpretation of the findings should consider certain methodological constraints. While the small sample size (n = 3) limits generalizability and precludes statistical inference, the application of MDC thresholds provided a structured framework for evaluating individual-level changes and contextualizing physiological outcomes. Additionally, although the session-specific biofeedback calibration introduced some individualization, the fixed-threshold algorithm may not have sufficiently challenged participants with higher baseline function or early neuromuscular gains. This could have limited training effects, particularly in those with greater functional reserve. Moreover, as discussed, the observed dissociation between improvements in tongue pressure and submental sEMG amplitude may reflect distinct underlying muscle recruitment patterns, which future studies should account for when selecting outcome measures. The potential influence of submental adiposity on signal quality, particularly in Subject 3, also highlights the importance of considering anatomical variability in future sEMG-based protocols. Lastly, while short-term feasibility and adherence were demonstrated, longer-term outcomes and sustained engagement—especially in the absence of perceived functional improvement—remain uncertain and warrant further investigation.

5. Conclusions

Taken together, these findings contribute to the growing evidence that mHealth-supported ES training can be a viable, accessible, and potentially preventive strategy for older adults at risk of dysphagia. The integration of real-time biofeedback appears to support both adherence and physiological adaptation. Future research should focus on optimizing training parameters, refining threshold-setting mechanisms, evaluating the synergy between muscular outcome measures, and validating functional gains through instrumental assessments. Given the sample size, the results of this study should be interpreted with caution. Larger, controlled studies with long-term follow-up will be essential to determine the durability of improvements and the potential of such programs to delay or prevent dysphagia onset in aging populations.

Author Contributions

Ali Barikroo: Funding acquisition, conceptualization, methodology, supervision, formal analysis, analysis, writing – original draft, writing – review & editing. Alyssa Zinser: Project administration, writing – review & editing. All authors have read and approved the published version of the manuscript.

Funding

This study was supported by start-up funding from Kent State University to PI Ali Barikroo.

Competing Interests

The authors have declared that no competing interests exist.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request. The data are not publicly available due to privacy or ethical restrictions.