This study is a secondary analysis of eMOM GDM RCT [20], where the intervention arm used the eMOM app with a CGM and activity bracelet for 1 week per month and a digital food diary for 3 days of this week (referred to as the ‘app week’ later in this paper), from gestational weeks 24-28 to delivery. The eMOM GDM study was conducted between February 2021 and December 2022 in Helsinki, Finland. Patients were recruited from antenatal clinics in the Helsinki metropolitan area by research nurses.

Participants eligible for the study were persons aged 18-45 years with a diagnosis of gestational diabetes at gestational weeks 24-28. Individuals were excluded from the study if they had preexisting type 1 or type 2 diabetes, were using medications that affect glucose metabolism (such as continuous corticosteroid therapy or metformin), had multiple pregnancies, physical disabilities, current substance abuse, severe psychiatric disorders that could complicate participation, or significant difficulties in cooperating, such as inadequate Finnish language skills.

Participants were randomized 1:1 into intervention and control arms. The intervention arm received standard clinical antenatal care in addition to the eMOM app, while the control arm received only standard clinical care. Standard clinical care consisted of periodic visits with a nurse and a doctor in municipal maternal clinics, as offered for all pregnant individuals in Finland. A regular care protocol for low-risk pregnancies includes 9-10 visits with a nurse and 2 visits with a doctor. After being diagnosed with GDM, individuals receive guidance on diet, physical activity, and self-monitoring of blood glucose with electronic capillary glucose meters [22].

The study was conducted in accordance with the Declaration of Helsinki and received approval from the ethics committee of Helsinki University Hospital (approval no. HUS/2165/2018). All participants enrolled voluntarily, provided written informed consent, and were informed of their right to withdraw from the study at any time without consequence. No financial compensation was offered for participation. All data were pseudonymized and nonidentifiable to protect participant confidentiality.

The eMOM mobile app was developed to visually represent data from a tissue CGM (Medtronic Guardian Connect with Enlite Sensor) and an activity bracelet (Garmin Vivosmart 3) together with a digital food diary. The information from the sensors and food diary is updated every 10 minutes to eMOM. In our previous study [23], physical activity was measured with 3 different devices (activity bracelet, hip-worn sensor, and electrocardiography sensor). Most participants preferred the bracelet to hip sensor, so it was selected as an activity measurement tool for this study. Screenshots from the eMOM app are presented in Figure 1. For displaying the data, the eMOM app has 2 main views: a day view (Figures 1A and 1D) and a week view (Figure 1E). Nutrition intake from each energy-yielding nutrient is visualized as stacked bars (Figure 1A). More detailed information (recorded food items and nutrient intake in grams) can be accessed (Figure 1B) by tapping the stacked bar (Figure 1A). The eMOM app shows detailed glucose values (Figure 1C) when tapping the glucose curve on the screen (Figures 1A and 1D).

The interactions with eMOM were collected in log files to determine how the participants used eMOM during the study. Each row in the log file included a pseudonymized user ID, timestamp, used view (day view, week view, detailed nutrition view, detailed glucose view, or info view), and filter used in day view (glucose, nutrition, and physical activity) or in week view (glucose, nutrition, physical activity, weight, and sleep).

The background questionnaire covered experience with self-tracking at baseline (gestational weeks 24-28). Other background information, such as age, BMI, parity, and previous GDM status, were collected from hospital registries.

Maternal learning and self-discovery were assessed through interviews conducted with randomly selected participants from the intervention arm. These interviews took place after participants had been using the app for approximately 2 months. Interviews were semistructured and carried out remotely over videoconferencing software. They were audio or video recorded depending on the participant’s preference. They lasted from 30 to 56 minutes; the mean duration was 40 (SD 8.5) minutes. The main interview questions were, “Have you learned something while using the eMOM app? If yes, what?” “Has the eMOM helped you to make changes in your lifestyle? If yes, how?” “What features of the eMOM were most useful for you?” “How would you develop the app?” The interview script is provided in Multimedia Appendix 1.

To assess maternal motivation, we used 2 questionnaires based on the self-determination theory: the Treatment Self-Regulation Questionnaire (TSRQ) and the Perceived Competence Scale (PCS) for diabetes [24]. The TSRQ questionnaire assesses levels of autonomous versus controlled motivation, while the PCS measures perceived competence. Both questionnaires were administered at baseline (gestational weeks 24-28) and again between gestational weeks 35-37. Details of the questionnaires and scoring methods are available in Multimedia Appendix 2.

The maternal depression was measured with the Edinburgh Postnatal Depression Scale [25] at baseline (gestational weeks 24-28) and again at gestational weeks 35-37. The scale consists of 10 items, each scored from 0 to 3, yielding a total score range of 0 to 30. A score of 10 or higher is commonly used as the threshold to indicate possible depression [25].

The acceptance of eMOM was measured with the Unified Theory of Acceptance and Use of Technology questionnaire [26] after each app week (once a month), and the usability of the eMOM app was measured once, 4 weeks after taking the app into use with a questionnaire adapted from the Software Usability Measurement Inventory questionnaire [27].

The Firstbeat Bodyguard 2, a chest-mounted sensor by Firstbeat Technologies, tracks heart rate variability (HRV) and physical activity using 3D acceleration over a continuous 3-day period. It uses 2 clinical-grade electrocardiographic electrodes for attachment and boasts a high accuracy in HRV measurement, with less than a 3-millisecond error and over 99.9% detection accuracy compared to standard electrocardiographic devices [28]. Additionally, Firstbeat’s Lifestyle Assessment software analyzes the HRV and motion data to evaluate stress and recovery levels, a methodology supported by extensive research validation [29,30]. Both the intervention and control arms used the Firstbeat sensor during the baseline period (gestational weeks 24-28) and again in the third trimester (gestational weeks 35-37). The sensor was worn continuously for 3 days during each phase.

The interviews underwent transcription, and 2 researchers (HY and SM) familiarized themselves with the content by reviewing the transcripts. The subsequent analysis followed the framework method, a recommended approach for multidisciplinary health research [31]. We used a combination of deductive and inductive approaches, where we had a set of initial codes (self-tracking of glucose, self-tracking of nutrition, and self-tracking of physical activity) at the beginning of the coding and used new codes when emerged from the gathered data. These new codes were combined into broader categories. The coding was performed by HY using ATLAS.ti software, which enabled charting of resulting themes and participants’ responses. This matrix presentation enabled us to analyze patterns, similarities, and differences in responses within each theme. Coding and the resulting themes were reviewed by 2 additional researchers (SM and MK). Quotes featured in the results were translated into English using an intelligent verbatim technique, which involves the removal of filler words like “er” during the translation process.

The log files were examined by computing the frequency and timing of interactions with the eMOM platform. Finally, we used triangulation across these various data sources (log files from eMOM, interviews, and questionnaires) to gain a comprehensive understanding of how self-tracking with eMOM supported self-discovery.

For comparing the difference between the arms, we performed a 2-sided t test for continuous outcomes, setting the significance threshold at 5%. For binary outcomes, we applied a chi-square test, resorting to the Fisher exact test when expected frequencies fell below 10. The difference between app weeks in the acceptance of eMOM was evaluated using repeated-measures ANOVA.

In analyzing the correlation between motivation and primary outcome from Kytö et al [21], we examined the linear relationship (Pearson correlation). For the binary outcomes, we used logistic regression.

In total, 148 participants with GDM participated in the study. Out of 76 persons in the intervention arm who used the eMOM app, 63 continued until delivery. The main reason for dropout was the need to start GDM medication. In total, 18 participants from the intervention arm were randomly selected for user experience interviews for this study. The log files were collected from all the participants in the intervention arm. Baseline characteristics are presented in Table 1. Participants were familiar with using mobile apps and activity trackers and had a good conception of the significance of different nutrients for the diet, as shown in Table 1. No significant differences were observed between the arms at the baseline.

Users interacted with the eMOM app most frequently during the morning hours (7-10 AM) and the evening hours (7-10 PM). Interaction levels were lowest at night (12-5 AM), increased sharply around 6 AM, peaked at 9 AM, and remained relatively steady throughout the day with a secondary rise in the evening.

During the first week of use, participants interacted with the app an average of 18.7 (95% CI 17-20.4) times per day. In the following application weeks, daily interactions ranged between 10 and 12 times. Between active usage weeks, the interaction frequency dropped significantly to 1-3 times per day.

Regarding app usage patterns, the day view was by far the most commonly used feature, accounting for 65% of total interactions (Table 2). This was followed by the week view (18%), detailed glucose view (10%), information section (4%), and detailed nutrition view (3%). When viewing data, users most frequently applied the glucose filter (65% of the time), followed by the nutrition and physical activity filters (each 15%), while the weight and sleep filters were used only 3% of the time (Table 3). Glucose was mostly viewed together with nutrition and physical activity, as shown in Figure 2.

The majority of interviewed participants (17/18, 94%) reported that the eMOM app facilitated their understanding of the connections between their daily activities and glucose levels. Primarily, participants found their learning experiences centered around nutrition (15/18, 83%) or physical activity (8/18, 44%). Seven (39%) participants acknowledged the combined influence of both nutrition and activity on their glucose levels. The interviews highlighted various methods of learning, such as experimentation, comparison, and observation.

Regarding self-discovery, 335 codes and 22 categories emerged from the interview data. These categories were then combined into main themes, which we present in a condensed form in the following sections. We have gathered all the supportive features of eMOM discussed in the interviews, along with direct comments from participants, and organized them into Table 4.

The majority of the participants (15/18, 83%) discovered the patterns of association between nutrition and glucose. Nutrition information was presented through stacked bar charts, with a possibility to be integrated with the glucose curve, enabling direct comparisons of how various meals impacted glucose levels. Participants engaged in experimentation and comparison, evaluating different foods and nutrients, mealtimes, and portion sizes.

As shown in Figure 2, glucose was mainly viewed alongside nutrition (43% of the time). Table 4 illustrates user comments on how nutrition data has contributed to their self-discovery.

Although the activity was viewed with similar frequency as nutrition (see Table 3) and nearly as frequently in parallel with glucose visualization (39% of the time; Figure 2), it contributed less to self-discovery. However, almost half of the participants (8/18, 44%) noted a decrease in their glucose levels following exercise. Participants also assessed their activity levels on different days to explore the relationship between their activity—both active and inactive days—and their glucose levels on those days. Table 4 summarizes user comments on how activity data contributed to their self-discovery process.

Sleep data were found to be challenging to interpret, with a mere 3% viewing frequency (Table 3). Four participants reported discrepancies between the sleep data provided by the device and their own assessments of sleep duration and quality. One participant observed an association between a poorly slept night and elevated morning glucose level.

The visualization of measurements emerged as a pivotal aspect of the learning process. Through observing the glucose curve, participants noticed fluctuations in glucose levels. When various filters (nutrition, activity, and sleep) were applied to the same graph alongside the glucose curve, participants could observe how different factors influenced glucose levels. The ability to view nutrition information in a quantitative format was also valued, and having a specific glucose target range was found to be important. Participants found the eMOM app more informative than Medtronic’s CGM in evaluating the overall picture of glucose balance. Table 4 summarizes the primary features used by individuals with GDM in the self-discovery process and how the eMOM app supported the identification of associations between these features and glucose levels.

In the evenings, eMOM served as a tool for obtaining a comprehensive overview of the day’s events, as 1 participant described the following.

Conversely, during the day, participants tended to rely on Medtronic for checking glucose levels, as it offered easier and quicker measurements due to its lower latency, updating every 5 minutes compared to eMOM's 10-minute intervals.

For most participants (n=10, 56%), the first week of using the app was found to be the most significant in terms of learning. Two participants expressed.

This reflects their strong interest in monitoring glucose levels during this period. As a result, interactions with eMOM were most frequent during this initial learning phase, as illustrated in Figure 2.

Periodic use of the eMOM app was found to be effective in maintaining motivation and sustaining interest in tracking activities. As 1 participant noted.

Some participants (5/18, 28%) expressed a desire for enhanced informativeness by having all three filters (glucose, nutrition, and activity) integrated into the same graph.

A desire for additional support in interpreting data was expressed by 11 (61%) participants, particularly in understanding the impact of sleep data on glucose levels. Moreover, personalized recommendations regarding nutrition and more specific daily goals for macronutrients were requested. Preferences regarding support varied. While 1 participant mentioned seeking guidance from a nurse to interpret glucose data and 2 participants wished for assistance from health care personnel, others (8/18, 44%) would have been satisfied with suggestions or notifications from the app.

In this intervention, the activity bracelet tracked only steps and overall daily activity. Participants desired a broader range of sports tracking capabilities. For example, swimming is a preferred activity among pregnant persons, but Vivosmart 3 was not able to track swimming. Issues with activity tracking, such as inaccurate step counting during activities like walking with a stroller or counting knitting as steps, led to a sense of distrust in the accuracy of activity data.

One participant expressed a desire for guidance on physical activity, including recommendations on duration and types of exercises to meet daily activity goals. Some also mentioned that motivation toward physical activity decreased as pregnancy progressed and exercise levels decreased.

Furthermore, some participants noted that when their glucose levels remained within target ranges, their motivation to use the app and adhere to a healthy diet diminished. As 1 participant articulated.

Autonomous motivation (measured with TSRQ) decreased in both arms: –0.11 (SD 0.38) in the control arm and –0.18 (SD 0.32) in the intervention arm between the baseline and gestational weeks 35-37. The between-arm difference was not statistically significant for the change of TSRQ (difference: –0.09; t₉₇=1.15; P=.25) nor for the change of perceived competence (difference: –0.27; t₉₇=1.21; P=.23). There was no statistically significant difference between the arms in the incidence of Edinburgh Postnatal Depression Scale scores greater than 10 (χ²₁=0.10; P=.75). The summary of the results on the motivation is presented in Table 5.

Correlations between autonomous motivation (TSRQ) and perceived competence (PCS) and improved primary and secondary clinical outcomes [21] in the intervention arm are shown in Table 6. No significant correlations were found in change of motivational factors and improved clinical outcomes.

The acceptance of eMOM and wearable sensors was high over the whole intervention (see Table 7 for a subset of statements and Multimedia Appendix 3 for responses to all the statements). For example, participants found the eMOM and the sensors useful for supporting lifestyle changes and they perceived eMOM easy to use.

Responses to the usability questionnaire (modified from the Software Usability Measurement Inventory) were in line with responses to the Unified Theory of Acceptance and Use of Technology. For example, 86% (38/44) of participants disagreed with the statement “Learning the functions of the eMOM GDM application takes too long,” and 66% (29/44) agreed with the statement “I understand and know how to act based on the information provided by the application.” 59% (26/44) of participants agreed with the statement “eMOM answers too slowly to inputs,” which may have influenced the preference for Medtronic, which updates more quickly, during the day and eMOM more frequently in the evening. See Figure 3 for all the results of the usability questionnaire.

We found that a mobile app combining CGM, nutrition, and physical activity within the same app was highly used and supported self-discovery among persons with gestational diabetes. Providing pregnant persons with the opportunity to compare the effects of nutrition and physical activity on glucose levels facilitated their understanding of cause and effect. Almost all participants using the mobile app discovered associations between either nutrition or physical activity, and glucose levels. The majority found associations between nutrition and glucose, and nearly half of the participants identified associations between physical activity and glucose. There were no differences between the intervention and control arms in the change of perceived competence to manage GDM, autonomous motivation, depression, or stress scores. Additionally, autonomous motivation did not influence the improved clinical outcomes observed in our previous RCT [21]. Acceptance and usability remained high throughout the entire study period [21].

The impact of an app that comprehensively integrates data in a single platform (including the automatic collection of activity and continuous glucose data) on self-discovery has not been previously studied in the context of gestational diabetes. Our study demonstrates that participants gained valuable insights into the associations between nutrition and glucose levels, facilitated by the eMOM app’s integrated data display. This comprehensive platform, which automatically collects activity and continuous glucose data, enabled users to observe the real-time impacts of their dietary choices on glucose levels. The ability to view glucose curves alongside nutritional details in a single window simplified the interpretation of meal timing and its effects, helping users better regulate their meal rhythms. Participants found value in comparing different foods and portion sizes, highlighting the importance of detailed nutritional information over simple carbohydrate tracking. These findings align with our previous work with various sensors, but without the eMOM app [23], where participants noted causal relationships between nutrition and glucose levels.

Learning the associations between physical activity and glucose proved to be somewhat more challenging than understanding the associations between nutrition and glucose. However, in this study, the associations between physical activity and glucose were more prominent compared to our study without the eMOM app [23]. The app’s parallel visualization of activity and glucose data may explain this improvement, as users could more easily observe associations between their physical activity and glucose changes. However, selecting features related to physical activity is challenging during pregnancy, as it complicates activity tracking. Activity bracelets typically track activity based on heart rate, which physiologically increases during pregnancy [32]. Currently, there are no activity bracelets designed specifically for pregnancy, which would account for these physical changes. While walking is a common form of physical activity during pregnancy [33,34], it may become difficult in the later stages. Swimming, water running, and weightlifting are preferred alternatives among pregnant individuals [33]. Therefore, it is essential to be able to track these alternative forms of physical activity. It is also worth noting that as pregnancy progresses, there is a general decline in physical activity levels [35,36]. Consequently, the relevance of physical activity in the management of GDM diminishes as pregnancy approaches its later stages. Similar challenges in physical activity tracking emerged in a study by Årsand et al [37] on type 2 diabetes, where patients expressed a desire for a broader range of trackable sports beyond just counting steps.

The eMOM app also provided data on sleep patterns, although interpreting this information proved challenging. Only 1 participant reported observing a potential connection between sleep and glucose levels, noting that during the flu accompanied by poor sleep, her glucose levels were higher. However, this association remains uncertain, as the flu itself can affect glucose levels. These results highlight the difficulty in finding associations between sleep patterns and glucose levels. Additionally, there is limited evidence on the effectiveness of wearable sensors on sleep during pregnancy [38]. Further investigation is needed in this area.

Recent research has explored ways to enhance the use of self-tracking data in diabetes self-management, although focusing primarily on type 1 and 2 diabetes [18,39-43]. However, self-management support is also crucial for individuals diagnosed with GDM, especially immediately following diagnosis, when adapting to the condition can be challenging [44,45]. Many individuals with GDM rely on trial-and-error learning [44-48], as seen in studies like that of Ibrahim et al [44], where participants expressed a desire to have all their self-monitoring tools gathered in one place to aid in sense-making. The recent review by Safiee et al [49] further supports this, showing that telemedicine interventions in GDM management improved self-management skills, particularly around understanding nutrition's impact on glucose. However, all prior reviewed interventions lacked CGM. Capillary glucose measurements capture only single-point readings, missing the full-day glucose fluctuations, including nocturnal hyperglycemia, which has been linked to a higher likelihood of large-for-gestational-age babies [50]. In contrast, the eMOM app provided a more comprehensive view of glucose levels, allowing participants to track changes over time. It is, however, worth noting that although CGMs have been found to be acceptable among persons with GDM [51-54], their use alone may not lead to improved glycemic control [54,55] or a reduction in macrosomia (birth weight > 4 kg) [53]. One contributing factor is the lack of clarity regarding the cause-and-effect relationships between lifestyle choices and glucose levels among persons diagnosed with GDM [45,47,56-58]. eMOM provides a tool for observing these causal relationships, thus enhancing the self-discovery process in managing GDM.

These findings emphasize the value of comprehensive data visualization tools in facilitating meaningful self-discovery in GDM management, offering both immediate feedback and clearer connections between daily behaviors and glucose outcomes.

While previous studies indicate a strong motivation among individuals with GDM to manage their condition [45,48,59], our study aimed to understand how eMOM influences motivation, with a particular focus on autonomous motivation and perceived competence—both key components of self-determination theory. Contrary to the findings of Williams et al [60], who demonstrated a significant correlation between improved glucose levels and autonomous motivation among type 2 diabetes patients, our findings suggest that the eMOM app did not lead to improved clinical outcomes through increased autonomous motivation. At the start of the intervention, our study participants in both the intervention and control arms were already highly motivated, leaving little room for further increases in autonomous motivation. Additionally, those using eMOM did not increase their perceived competence to manage GDM compared with the control arm. These findings suggest that the improved maternal clinical outcomes reported earlier [21] are primarily attributed to self-discovery rather than increased motivation. Thus, motivational interventions that are effective in type 2 diabetes management may not directly apply to GDM. Instead, educational approaches, such as the eMOM app, play a crucial role in helping persons with GDM learn how to manage their condition effectively.

This study is the first to combine qualitative and quantitative approaches to investigate how comprehensive self-tracking influences self-discovery and motivation in managing GDM. The eMOM app offers a novel approach to GDM self-management by integrating CGM, nutrition tracking, and physical activity monitoring in real-time within a single app, without guidance from health care personnel, thus addressing a gap in previous GDM apps [61]. This integrated approach was also requested in our previous study, where participants used the same variable sensors and meters (CGM, physical activity, food diary, and sleep) as in this study, but separately in different apps and devices [23].

While some previous GDM apps have aimed to reduce reliance on health care personnel [15,16,62,63], none have incorporated the level of comprehensive and automated data tracking offered by the eMOM app. To our knowledge, only 1 earlier app, Habits-GDM by Yew et al [15], combined glucose data, a food diary, and physical activity tracking. It used capillary glucose measurements and monitored physical activity solely through step counts using a phone pedometer. The app also displayed only total calories and carbs, lacking detailed nutritional data. While Habits-GDM resulted in lower proportions of glucose levels above target and reduction in composed neonatal outcomes, none of the individual neonatal outcomes were statistically significant, and the intervention did not reduce gestational weight gain. Habits-GDM also included a messaging feature, allowing users to consult health care personnel and receive responses within 24 hours. However, the study did not report information on the app’s usage rates.

Similarly, MobiGuide by Peleg et al [62] combined capillary glucose measurements with step counts but lacked nutrition tracking. While it showed a significant reduction in blood pressure, it did not demonstrate any significant differences in glucose levels, insulin use, or neonatal outcomes. The Pregnant+ app by Borgen et al [63] allowed users to manually log physical activity, but none did, highlighting that manual logging alone was insufficient for meeting users’ needs.

As reported in the RCT study [21], eMOM use was consistently high. Most of the learning occurred, however, during the first application week. In the following application weeks, eMOM was mainly used to check if the glucose stayed within the limits. As noted by the participants, continuous availability of the sensors might lead to decreased motivation for tracking. Due to this periodicity, engagement with eMOM remained high during the application weeks, in contrast to the declining engagement observed in automatic self-tracking among people with type 2 and type 1 diabetes [64]. Participants continued to interact with eMOM even outside the designated application weeks, even when they did not have access to the activity bracelet or CGM. However, these interactions were significantly less compared to the application weeks. Throughout this study, the acceptance of eMOM and wearable sensors remained high, as indicated by questionnaires.

Previous research has suggested that extensive self-monitoring could lead to adverse psychological outcomes [65,66]. We found no evidence of an effect on depression or stress, as measured by Edinburgh Postnatal Depression Scale questionnaires and Firstbeat stress scores. This suggests that comprehensive tracking does not have a detrimental effect on maternal psychological well-being.

The most significant limitation concerning the self-discovery process was the inability to display nutrition, physical activity, and glucose data together in 1 graph. Participants expressed a need for more support in interpreting data and making informed choices. For example, the app could send notifications when glucose values exceed the target range and instruct users to review their recent meals.

Self-discovery’s impact on clinical outcomes was not objectively measured. However, the results of the interviews suggest that self-discovery had a positive effect on clinical outcomes. Another limitation is that individuals who required medication for GDM were excluded from the study. This may have introduced bias into the results of motivation, depression, and stress levels, as the analyses only included patients whose condition was managed well enough to avoid medication.

Implementing digital tools in clinical care requires patients to have health literacy. Low health literacy in patients with type 2 diabetes has been linked to lower engagement with mobile-based health interventions [67]. To be effective, information provided by these apps should be tailored to the health literacy levels of the target population [68]. In future research, eMOM should be evaluated in populations with varying levels of health literacy. Artificial intelligence–based solutions could assist users in understanding lifestyle improvements based on their health data.

Our findings suggest that an app presenting features in a quantifiable format is well-suited for the self-discovery process in GDM. Additionally, the limited number of features in GDM and the clear causal relationships between glucose and these features make this process easier to learn. Self-discovery in GDM exemplifies goal-driven tracking [69], particularly for monitoring nutrition and physical activity to maintain glucose levels within the recommended limits set by health care providers. We believe that our approach of integrating data from various sources in real-time could be beneficial for other noncommunicable diseases, such as type 2 diabetes, where the relationships between features and outcomes are relatively clear and learnable. Especially after diagnosis, when individuals are still in the learning phase of managing their condition, patients could benefit from a solution like ours.

Beyond the short-term effects on pregnancy outcomes, gestational diabetes also has long-lasting implications for maternal health [2,3]. Future research should investigate the extended impact of using the eMOM app, particularly in the postpartum period. Examining how this tool influences long-term health behaviors, psychological well-being, and the prevention of type 2 diabetes could offer valuable insights into its role in promoting sustained self-management beyond pregnancy.

By visualizing CGM, nutrition, and activity in a single app in real-time, the eMOM app provides a promising tool to enhance self-discovery in GDM, without additional support from health care personnel. Our findings also suggest that eMOM’s impact is more related to self-discovery than to increased motivation. To support learning further, future work involves visualization of more than 1 feature impacting glucose at the same time and the use of artificial intelligence–based recommendations. In addition to GDM, this kind of mobile health solution could be used to support learning in other noncommunicable diseases, such as type 2 diabetes.