The telemedical platform for patients with MG, “MyaLink” used in the study was developed at Charité Universitätsmedizin Berlin together with software partner Qurasoft GmbH (based on the SaniQ platform). The platform is a Conformité Européenne (CE)-marked class-I medical device (Medical Device Directive) and is GDPR (General Data Protection Regulation) and BlnDSG (Berlin Data Protection Act) compliant. The platform consists of an app for patients (iOS and Android) and a web-based portal for physicians facilitating to review patient data, manage and set up individual monitoring plans and medication plans. It offers a secure interactive platform for exchange of information and health data between physicians and patients and informed treatment decisions (Figure 1). The MyaLink app contains the following modules: (1) transmission and integration of sensor-based data from wearables including heart rate, oxygen saturation, and step count (device: Garmin Vivosmart 5); (2) assessment of the forced vital capacity from a digital spirometer (device: MIR Spirobank Smart One); (3) assessment of questionnaires and PROMs (see details under Section 3.5); and a (4) communication module for individual requests and patient-physician interaction. Furthermore, the app contains a (5) medication plan with reminder function and (6) a module including data storage of important documents (eg, clinical reports), which can also be shared with the physician through the web-based portal. The web-based system for physicians enables real time access to this data and allows clinicians to obtain an accurate overview of the patients’ condition and visualization of health data. If necessary, the patient can be contacted through the communication tool and adjustments to medication and the remote monitoring plans can be made.

Inclusion and exclusion criteria have been previously presented in detail in our methods paper [17]. Inclusion criteria included patients with a diagnosis of MG for at least 6 months before study enrollment, aged 18 years and older, and classified according to the Myasthenia Gravis Foundation of America (MGFA) classification status I-IV [18]. Exclusion criteria included patients with somatic, psychiatric, or neurological disorders who were undergoing treatments or were not stabilized under drug therapy that could adversely affect cognitive function. Patients were recruited through the specialized outpatient myasthenia center at Charité Universitätsmedizin Berlin.

This analysis focused on data obtained from the telemedical intervention group from a randomized controlled trial (RCT) conducted at our specialized center at Charité Universitätsmedizin Berlin in Germany. The RCT included 45 patients with MG over a 12-week observation period between April and September 2023. Patients were randomized in a 2:1 ratio into the telemedical intervention group (n=30) or the control group (n=15), who received standard of care without telemedical intervention. Randomization was performed according to gender and clinical class stratified by MGFA score at baseline (using the randomizer tool [19]). The primary objective of the study was to investigate the feasibility of the remote monitoring platform in care of patients with MG. Feasibility was measured through adherence rates based on completion of PROMs in the app and data collection from external devices (including a digital spirometer and activity tracker). Furthermore, user satisfaction with the telemedical platform was assessed among patients and health care professionals comprised of the study team (n=5) at the end of the study. We, therefore, exclusively analyzed data from the 30 patients from the intervention group. Secondary objectives of the study (but not part of this analysis) included comparing standard of care with and without additional telemedical intervention. This comparison focused on evaluating clinical endpoints between the control and intervention group. Comprehensive baseline and end-of-study visits were conducted at the study center for both groups and included an in-person clinical assessment of medical history, medication, hospitalizations for MG, care-related questions (eg, accessibility of a specialist or areas of improvement regarding MG treatment), various PROMs and MG-specific clinical assessment through the quantitative myasthenia gravis score (QMGS) [20]. At baseline, the intervention group received a wearable (activity tracker) and a digital spirometer, along with the MyaLink app, which were set up and connected on the patients’ smartphone. Instructions on app and external devices usage were provided by the study team. This enabled remote health monitoring throughout the study by active (PROMs and spirometry) and passive (wearable) continuous data collection. Patients were instructed to fill out PROMs and perform spirometry at predefined intervals. They were instructed to wear the wearable during waking hours (at least 14 h per day) throughout the study period and optionally at night. Telemedical check-ups occurred in weeks 4 and 8, comprising the reviewing of remote monitoring data by study physicians and contacting the patients if necessary (eg, changing medication). Patients could also initiate communication through the chat module at any time, with physicians monitoring messages on weekdays during working hours and responding within 24 hours. Patients completed usability questionnaires through the app a week before the end of the study, while health care professionals conducted their questionnaires after study completion. Furthermore, oral feedback from patients was collected and documented during the end-of-study visit. All data from the study visits and all data from the telemedical check-up was documented in REDCap (Research Electronic Data Capture) software (version 13.7.31; Vanderbilt University).

In total, 6 questionnaires and 1 test were assigned to patients through the app, administered weekly or monthly, depending on their specific type (Figure 2). Weekly questionnaires included the myasthenia gravis activities of daily living (MG-ADL) profile [21], myasthenia gravis quality of life, revised (MG-QoL15r) [22], and single simple question (SSQ) [23], which are commonly used in MG studies including large approval studies to assess functional status and quality of life. Patients were also instructed to perform a weekly breath count test [24]. Monthly questionnaires included Chalder Fatigue Scale (CFS 11) [25] and Hospital Anxiety and Depression Scale (HADS) [26] to assess common comorbidities in MG. In addition, Patient Acceptable Symptom State (PASS) was assessed monthly [27]. Furthermore, patients completed 2 user satisfaction questionnaires at the end of the study. For broad comparability, patients completed the System Usability Scale (SUS) [28], where patients ranked each of the 10 statements from 1 to 5 based on how much they agreed with the statement. The second questionnaire was adapted from Braun et al [29] and specifically tailored to the MyaLink telemedical platform. It features 17 items categorized in 3 sections and uses a 4-point Likert scale (strongly agree to strongly disagree). Originally formulated in German for publication purposes, it underwent consensus translation into English by 4 medical health care professionals, including 1 native speaker. After the study, the study team completed a modified version of the questionnaire adapted from Braun et al for health care professionals, respectively.

Patients were instructed to perform spirometry measurements weekly, with reminders sent through the app through push notifications. The wearable continuously collected oxygen saturation level (every minute), heart rate (every minute), and steps (hourly aggregated). Data from both the digital spirometer and the wearable were transmitted directly to the app by bluetooth. The study team regularly checked the web-based system to ensure receipt of data from external devices and questionnaires, sending reminder messages through the chat module when necessary.

The statistical calculations were performed using R software (version 4.2.2; R Foundation for Statistical Computing) [30]. Descriptive statistics (median [IQR] and absolute and relative frequencies) are presented depending on the scale of the variables.

Feasibility was assessed by adherence rates. For PROMs, assignments of the questionnaire started a week or month after the baseline visit respectively, depending on the questionnaire. Questionnaires were assigned together with a push notification on the patient’s phone and disappeared from the app upon completion or were “overwritten” upon receiving a new assignment. Questionnaires could only be submitted by patients through the app if they were fully completed. Each questionnaire was required to be completed within 7 days to be counted as “completed,” including the monthly assigned questionnaires, with missing entries calculated accordingly. The final day for data acquisition from the last assigned weekly questionnaires ended the day before the end-of-study visit. Total adherence rates were calculated as the total number of completed questionnaires divided by the total number of assigned questionnaires. Additional questionnaire assignments were allowed in cases of clinical deterioration but were excluded from adherence calculations. The study was considered successful if at least 60% of assigned PROMs through the app were answered by patients of the intervention group. Across the 30 patients of the intervention group, a mean adherence rate and a 2-sided 95% CI are determined. If the observed average adherence is at least 76%, with data from 30 patients, the 95% CI would range from 60.7% to 91.3%, thus exceeding the 60% threshold.

Spirometry adherence rates were calculated accordingly (ratio of the total number of actual measurements performed during the study vs total number of measurements required per the study protocol).

Adherence calculations for wearable data included measurements starting the day after the baseline visit and ending the day before the end-of-study visit. Adherence to wearable data from the activity tracker was evaluated by aggregating data over hourly windows, with 100% adherence defined as wearing the device for 14 hours per day per patient (based on the estimated waking hours). This was then extrapolated over the study duration. Each hour necessitated at least one heart rate measurement to validate wear time, considering heart rate a more reliable marker than oxygen saturation. Overall adherence was determined by the sum of the hours the patients had with at least one heart rate measurement divided by sum of the hours per day corresponding to 100% individual adherence.

The user satisfaction questionnaire was descriptively evaluated using the Likert scale. The SUS ranked each of the 10 statements from 1 to 5 based on user agreement, from which a SUS score was calculated: Each patient responses undergo conversion into a new numerical value (0‐4), which is then aggregated and multiplied by 2.5 to transform the original scoring range of 0‐40 into a 0‐100 scale. These scores do not represent percentages but a percentile ranking [31]. Research suggests that a SUS score above 68 is considered acceptable, with scores below 68 classified as not acceptable. Therefore, a score above 68 on the SUS is interpreted as a positive result further supporting the platform’s usability. A more detailed classification considers SUS scores between 73 and 84 as good, between 85 and 99 as excellent, and 100 as the best.

This analysis focuses on the primary study endpoints assessing feasibility and usability. Further exploratory analyses of secondary endpoints, eg, comparative clinical endpoint analyses (between control group and intervention group), evaluations of communication patterns from the MyaLink chat and telemedical interventions are currently prepared and results will be presented in further studies.

This study received ethics approval by the ethics committee at Charité Universitätsmedizin Berlin (EA2/157/22). Written informed consent was obtained from the study participants before the start of the study. The study was conducted in accordance with the Declaration of Helsinki (2013) and its later amendments. Data storage and handling responsibilities were approved by the ethics committee of Charité Universitätsmedizin Berlin. The data collected through MyaLink is stored on certified data servers, which adhere to a management system according to ISO/IEC 27001:2013.

Patients were between 23 and 82 years (median 47.5 years; IQR 40.5-57.5) at time of enrollment. The majority of patients were female (21/30, 70%). Most patients in the study had a mild to moderate disease severity, with 36.7% (11/30) categorized in MGFA II (a or b) status and 43.3% (13/30) in MGFA III (a or b) status at the time of enrolment. Median scores for MG-ADL and QMG were 10 (IQR 5.25-11) and 15.5 points (IQR 12.25-20.75), respectively. On average, patients were diagnosed 5.5 years (IQR 2-9.75) before entry in the study. One patient withdrew from the study due to health reasons unrelated to MG. All results presented include data from the dropout, including measurements taken before the time of dropout (with the patient’s consent). Detailed patient characteristics are presented in Table 1.

The total adherence rates for PROMs and external devices data ranged from 74.3%- 97.9% (1830/2464 - 327/334) across all data types and are detailed in Table 2. When examining adherence by data category, adherence to PROMs was the highest, followed by spirometry, and then wearable data from the activity tracker.

Adherence to PROMs ranged between 93.2% (55/59) and 97.9% (327/334), depending on the specific questionnaire. Patients were assigned weekly questionnaires and spirometry measurements between 11 and 13 times (drop out: 5 times) during the study, depending on individual study duration of a participant. Monthly questionnaires were assigned twice to all patients (drop out: 1). There was no difference in adherence regarding the questionnaire frequency. For the breath count test, adherence was 97.6% (326/334). Adherence to spirometry was slightly lower than PROMs (293/334, 87.7%), likely due to the higher task burden associated with these active measurements (Figure 3). Patients conducted a median of 49 spirometry measurements (IQR 15-59).

On average, patients wore the wearable for a median duration of 20 hours per day (IQR 15‐24). Wearable adherence rates exhibited considerable variability among patients, ranging from 6.8% (4/59) to 100% (83/83) (Figure 4), despite an overall strong adherence of 80.9% (1830/2261) for all patients during the study when excluding days without any data input (no recorded heart rate measurements) from the calculation (Table 2). Including days without wearable data input to the calculation decreased adherence to 74.3% (1830/2464). Clinical insights from chat interactions, as evidenced by the timestamps within the chat messages, unveiled several factors impacting patients with decreased device use. For example, out of the 30 patients, 2 patients (6.7%) experienced a myasthenic crisis necessitating prolonged hospitalization and reported having forgotten the devices at home. Furthermore, 1 patient (3.3%) had difficulty with using the app (older patient who was overwhelmed by the technical operation and conducting measurements with the digital spirometer) with numerous interactions with the software provider’s support team. Some patients experienced discomfort with the rubber strap of the activity tracker. Some patients had device-related issues such as (recurring) pairing problems and 1 out of the 30 patients (3.3%) received a defective activity tracker, necessitating a device replacement. However, even after replacement the patient reported ongoing connectivity issues between the wearable and his smartphone.

The highest levels of agreement were observed in the following order: experience with the care program, communication, and evaluation of the program. Regarding the care program evaluation, 89.7% (26/29) agreed that the app was easy to use while 93.1% (27/29) agreed that the time investment required for app usage was appropriate. In total, 89.6% (26/29) of patients agreed that describing their health status during follow-up appointments was easier. A total of 93.1% (27/29) indicated that treating physicians were interested in reviewing the collected monitoring parameters. In addition, 79.3% (23/29) of patients agreed that their quality of life had significantly improved since app usage. Futhermore, 96.6% (28/29) expressed willingness to recommend the app to other patients. In addition, 69% (20/29) agreed that they felt more confident in identifying MG-specific problems using the app. Views on the reduction of visits to neurologists or clinics were mixed; 55.2% (16/29) agreed to needing fewer visits to neurologists, while 51.7% (15/29) agreed to reduced visits to clinics, with 13.8% (4/29) strongly disagreeing with both categories (Figure 5). Out of the 29 patients, 14 patients (46.7%) responded that they did not agree that they required less frequent hospital visits; of those 14 patients, 8 patients (57.1%) were hospitalized during the study period due to MG (two of whom were in intensive care). In addition, 7 of these 14 patients (50%) experienced at least one exacerbation. Results of user evaluation from the health care professionals showed similar high acceptance rates regarding the telemedical platform, ease of use, and overall positive evaluation on the effect on patient care (Figure 6).

The overall median SUS score was 85 (IQR 75-90). Figure 7 shows that there was a high level of agreement regarding the system’s ease of use (26/29, 90%), patients’ confidence in operating it (25/29, 86%), and patients’ belief that others would quickly learn to use the app (25/29, 86%). Notably, 21% (6/29) remained neutral regarding perceived inconsistencies in the app, although the majority (20/29, 69%) tended to disagree. Among patients (N=22) who responded at least once with a neutral (Likert-scale 3) or negative response to any question (agreeing or strongly agreeing with negatively formulated questions or disagreeing or strongly disagreeing with positively formulated questions), there were 14 females (63.6%) with a median age of 47.5 (IQR 39.25-61.25; mean 49.7, SD 15.70), and the MG-ADL score at baseline was median 9.5 (IQR 5-11; mean 8.5, SD 3.60). Of all 29 patients, 69% (20/29) expressed willingness to use the app regularly. Among the patients who responded negatively or neutral to this statement, the documented oral feedback collected at the end-of-study visit mainly encompassed technical feedback, such as suggestions for adjusting the size of the chat window, implementing filter options for monitoring parameters in the health diary, and presenting preferences for the parameter dashboard. Furthermore, patients also frequently noted necessary improvements for a more user-friendly medication plan. In addition, there were critiques regarding the omission of specific symptoms in the PROMs (eg, nonmotor symptoms, pain, and sleep) and requests from patients for a free-form diary function (eg, for documenting infections or daily fluctuations).

Our study encountered several challenges associated with remote monitoring technology: These included missing passive wearable data primarily due to technical issues and synchronization problems and patients’ discomfort with the wristband from the wearable. Addressing these challenges and providing solutions, such as training sessions and available technical support, is crucial to ensure optimal patient experience and adherence in future studies using these technologies. Furthermore, the study cohort is relatively small, and the study duration is short. Patients with higher digital literacy may have been more likely to participate, potentially impacting the representativeness of the cohort. Regarding user evaluation of health care professionals, it should be noted that the study team size was small and partially comprised members of the development team for the platform, which could add to bias in answering the questionnaire.

In conclusion, our results demonstrate the feasibility of remote monitoring for collecting both subjective (active) and objective (passive) data from multiple sources among patients with MG, as evidenced by above-average adherence rates and acceptability. This underscores the potential for user-friendly digital tools for long-term monitoring of patients with MG and their willingness to track disease progression. Larger and prolonged studies are needed to benchmark our findings and to further explore the effective implementation of such tools and their clinical implications in practice.