The selection process of the apps was conducted following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines [21]. Between September 30 and October 5, 2022, we conducted searches on both the Google Play Store and iOS App Store using Korean keywords, including “mental health,” “mental health care,” “depression,” and “stress.” These keywords were chosen based on previous studies related to mental health and mental health apps. Following this initial search, we established specific criteria for app inclusion, considering prior research findings [22-24]. As free apps are generally more accessible to users and tend to be the most widely used, free downloads were included as a selection criterion [15]. The inclusion criteria for apps selected for the final analysis required that all of the following conditions (C1-C7) be satisfied:

The star ratings for each app were extracted from the respective app stores as of the search date. On both platforms, star ratings are based on a 5-point scale. The review counts represent the cumulative total number of user reviews recorded for each app up to the search date.

We downloaded all selected research apps and categorized their main contents in Multimedia Appendix 1, classifying them into categories, such as information provision, self-diagnosis, and monitoring functions. A frequency analysis was performed to determine the number of apps offering these functions.

For each of the final selected apps, we assessed quality using the Korean-translated version of the user version of the Mobile Application Rating Scale (uMARS) [26]. uMARS is a 5-point scale developed from the original Mobile Application Rating Scale to enable users to evaluate mHealth apps [27,28]. The scale demonstrates excellent internal consistency and has been widely validated, confirming its reliability and validity. The scale has been successfully applied to assess various mHealth apps, including those related to weight management, rheumatic conditions, menstrual cycles, and cancer risk calculators [22,29-31]. The overall uMARS score was calculated by averaging the scores across all 5 categories: engagement, aesthetics, functionality, information, and subjective quality.

Between October 8 and October 14, 2022, a total of 7 users of real mental health apps participated in the uMARS evaluation. The number of evaluators was determined based on user-centered evaluation standards, which suggest that 5 or more users are typically sufficient to identify most usability issues. This sample size ensures diversity in user experiences and allows for reliable results [32,33]. Apps were randomly assigned to evaluators to minimize potential bias due to subjectivity. Each evaluator was instructed to use the assigned app for at least 10 minutes daily, and evaluations were conducted in a blind test format.

After normalizing the data, the correlations among app content, uMARS scores, star ratings, and number of reviews were analyzed using Pearson correlation analysis. To explore the relationship between objective and subjective indicators, 4 separate correlation analyses were conducted by categorizing the aforesaid data into all apps, the top 20 apps, Android apps, and iOS apps. Multiple regression analysis was performed to identify the content factors influencing uMARS scores and evaluation items. All analyses were conducted using R software (version 4.1.2; R Core Team).

A total of 896 apps were identified using the search keywords. Apps that did not meet the inclusion criteria (C1-C7) were excluded, resulting in 41 apps being selected for the final analysis. This included 26 Android apps and 15 iOS apps (Figure 1).

The app’s operating system (OS) is linked to the App Store, allowing app updates whenever the OS is updated [34]. Therefore, depending on the timing of the OS update, the version and function of each OS may differ, and the same app on different OS versions was considered a different app.

Of the 41 apps, 29 (71%) included reminders, as well as recording and statistics features. Among the monitoring functions, 24 (59%) offered diary features, and 21 (51%) provided emotion tracking features. In terms of privacy, 26 (63%) included a lock feature, which was more common than in existing mHealth apps. However, only 1 (2%) provided an icon change option (Table 1).

An evaluation of the quality using uMARS resulted in an overall mean score of 3.53 (SD 0.34). Among the apps, “Mind Cafe Lite” for Android achieved the highest score of 4.08 (SD 0.30), while “Depression Test” received the lowest score of 2.68 (SD 0.57). In the uMARS evaluation, the top 5 apps received their highest scores in the information category. They shared a common feature: providing information and counseling services from counselors and counseling agencies. However, most of these services, including counseling and meditation, required a fee.

Conversely, the bottom 5 apps received the lowest scores on subjective evaluation items. These apps offered limited or no counseling-related services commonly provided by the top 5 apps. Moreover, most of them offered only a single function, such as stress measurement or depression testing, which limited their utility for consistent mental health management and promotion (Table 2). uMARS scores of all 41 apps are presented in Multimedia Appendix 2.

The top 5 apps commonly provided information about health care providers, while 3 of the apps included content related to mental health and cognitive behavioral therapy. One app offered counseling and meditation services, but these were available only through paid features such as in-app purchases. By contrast, among the bottom 5 apps, none provided information about health care providers, and only 1 offered mental health information. Most of the bottom 5 apps had single functions, such as stress measurement, depression testing, or emotional record-keeping, making them less suitable for ongoing mental health management and promotion (Multimedia Appendix 3).

Table 3 shows the correlations among the content of mental health apps, uMARS scores, star ratings, and number of reviews. Health care provider information (r=0.53; P<.001) and counseling/question and answer (Q&A; r=0.55; P<.001) had the strongest correlations with the uMARS score. Mindfulness and meditation (r=0.40; P=.01) and chatting algorithms (r=0.34; P=.03) showed weak correlations with it. Star rating was not correlated with app content. Recording and statistics (r=0.41; P=.009) and community and social networking site sharing (r=0.33; P=.04) showed weak correlations with the number of reviews.

Figure 2 depicts the correlation between users’ objective uMARS score, subjective star ratings, and number of reviews. Among all apps (r=0.51; P<.001) and Android apps (r=0.61; P<.001), there was a significant correlation between the uMARS scores and star ratings. Meanwhile, no significant correlation was found between uMARS score and star ratings among the top 20 apps (r=–0.08; P=.74) and iOS apps (r=–0.04; P=.89).

Table 4 presents the outcomes of the multiple regression analysis used to identify the factors influencing uMARS scores. A higher presence of information content in an app (β=.56; P<.001) and a higher star rating (β=.36; P=.02) were associated with higher uMARS scores. Furthermore, we examined the factors influencing each evaluation item within the uMARS assessment. The presence of information content significantly influenced the overall uMARS overall score. Moreover, apps with higher star ratings had correspondingly higher scores in the aesthetics category (β=.12; P<.001) and the subjective evaluation category (β=.34; P=.04; Table 5).

The development of various mental health apps is on the rise, and the use of mHealth apps is becoming increasingly vital in the mental health field. Nevertheless, few studies have assessed user satisfaction and app quality in the context of mental health apps. In this study, we used uMARS to evaluate mental health apps from the user perspective and to identify factors that influence user evaluations. Users preferred apps that provided information related to mental health promotion, including details about counselors and counseling institutions.

However, only a limited number of apps in our final selection provided this type of information. Mental health apps are effective in enhancing patient knowledge and symptom management, primarily due to their multimedia and audiovisual content, which can deliver mental health information and education [35,36]. Users generally perceive online health information as reliable, fostering trust in their treatment regimens and encouraging active participation in disease management [37]. Moreover, the inclusion of mental health information within these apps can contribute to improved mental health literacy among users [38]. To effectively support users in promoting their mental health through apps, future app development and updates should prioritize the inclusion of mental health–related content.

Counseling and meditation services, which constituted the primary content of the top-rated apps in user evaluations, were mostly provided for a fee. Although these apps can be downloaded for free, they often require payment for in-app purchases or access beyond a limited free trial period. The cost factor remains a prominent concern among mental health app users, and apps offering paid services may be less likely to be used [15]. Users generally prefer free apps that support independent management and monitoring of their mental health [39]. In Korea, public institutions are developing various mental health management apps; 58 of the 896 (6.47%) apps in our initial search were developed by public institutions. However, these apps were not included in our final analysis due to their extended update cycles, limited utilization, and small number of user reviews. Furthermore, mental health–related apps developed as part of research projects often focus on study objectives rather than user needs [35]. Evidence-based mental health promotion projects in the public sector are also relatively rare. To enhance app quality and encourage the use of public apps for mental health promotion, it is advisable to benchmark the services offered by top-ranking paid apps and align them with research-driven initiatives.

Chatting based on algorithms demonstrated a weak but significant correlation with uMARS scores. Users are increasingly interested in utilizing AI chatbots that provide convenient access to medical information without relying solely on health care providers [40]. Emerging technologies, such as generative AI, have the potential to significantly impact the health care sector by delivering tailored health advice, support, and integrated systems for detecting and managing health issues [8,41]. In the field of mental health, generative AI can serve as a valuable tool to support the treatment process by offering functions such as 24-hour online support, counseling record maintenance, and follow-up management [42]. It also has the potential to bridge gaps in the diagnosis and treatment of mental disorders, expanding access to psychiatric care [43]. Currently, AI chatbot functions in mental health apps are somewhat limited, often restricted to depression self-diagnosis tests (eg, the 9-item Patient Health Questionnaire). Additionally, technical issues such as limited response capabilities and unnatural conversation flow have been noted. The lack of personalized feedback for continuous engagement also hinders user satisfaction and long-term participation [44-47]. If AI counseling services that provide evidence-based, customized solutions by integrating existing mental health apps with generative AI functions are developed, they are expected to deliver higher-quality mental health services to users. To achieve this, it is essential to explore future app development strategies that incorporate generative AI in diverse ways. These efforts should focus on developing advanced algorithms that enhance natural conversation flow and provide sophisticated, personalized support. Furthermore, continuous clinical validation is necessary to ensure the reliability, accuracy, and safety of AI technologies.

The correlation between the uMARS score, an objective quality evaluation tool for mHealth apps, and user star ratings differed depending on the platform. Considering previous studies reporting digital disparities and age differences between Android and iOS users [48,49], these factors may have contributed to variations in app evaluation criteria, user preferences, and expectation levels. However, some studies have found no significant differences in personality traits or characteristics between users of the 2 platforms [50]. Combining these conflicting findings, it can be inferred that differences in user evaluations across platforms may be influenced not only by user characteristics but also by app usage experiences and platform-specific features. Future research should develop evaluation strategies that systematically analyze these differences and account for platform-specific characteristics.

Mental health apps collect and share sensitive information, including user symptoms and counseling records. When handling such sensitive personal data, it is crucial to build trust with users by ensuring the protection of their personal information [51]. Many users consider features such as app locking to be essential when using mental health apps [39]; however, few of the apps included in our final study offered the option to change the app icon for privacy purposes, and some even lacked lock functionality. Moreover, concerns about unclear personal information protection, the absence of relevant regulations, and a lack of personalized attention from apps have been steadily raised in mHealth research. As a result, some countries are now beginning to implement personal data protection regulations [52-54]. While regulations such as the Health Insurance Portability and Accountability Act and the General Data Protection Regulation were developed to protect personal information, they are not universally applied, leaving privacy and security challenges unresolved [55]. In response to these challenges, researchers are actively exploring sustainable systems to safeguard personal information using blockchain technology [56]. Blockchain-based health care systems offer personal information protection and provide users the option to request the deletion of their data [57]. In the mental health field, preserving the security and confidentiality of users’ mental health data is considered paramount [58]. Consequently, app developers must prioritize including privacy features suited to the app’s characteristics and aligned with existing personal information protection regulations.

None of the apps in our final analysis targeted specific population groups. Although apps for children and adolescents are in development, they remain difficult to find, and their specific benefits for particular target groups are still unclear [59]. Mental health issues vary across life stages, implying that desired app content may also differ by age group [60]. Previous research highlights the importance of tailoring interventions to life stages due to variations in mental health problems [61,62]. However, such a life stage–specific approach has yet to be fully integrated into app development. Future efforts should focus on developing mental health apps that address the unique needs of each life stage, considering factors such as sex and age. This requires active research into the usability and effectiveness of apps across different population groups to better address the diverse needs of users.

The World Health Organization (WHO) and the Centers for Disease Control and Prevention (CDC) offer targeted guidelines for mental health management, categorized by different population groups such as parents, adults, children, individuals with mental health conditions, public health workers, and health care professionals [63,64]. WHO also provides translated stress-management guides for use in different countries, along with illustrated materials that are easy for users to follow [65]. The CDC has created a dedicated page that consolidates guidelines from various institutions and includes links to mental health experts that can be accessed during emergencies. This allows users to access free or low-cost mental health services through a structured system based on standardized criteria, enabling them to choose an expert, appointment time, and office environment. By contrast, such detailed interventions are not yet standard practice for mental health management in Korea. Consequently, mental health services are often underutilized. To promote mental health services in the future, concerted efforts are required to raise awareness and ensure easy access to services through diverse interventions, such as national guidelines and expert-led initiatives offered by the WHO and CDC.

This study had some limitations. First, the apps were selected based on searches conducted within app stores during a specific period. As app stores are continuously updated, the content analyzed in this study may differ from that of future app versions. Second, the data sources were not electronic academic databases but app stores, and the accuracy of the search process may have varied depending on each store’s recommendation algorithms.

In this study, users evaluated the quality of mental health apps using a validated scale. Users preferred apps that included information related to mental health. To develop high-quality mental health apps, it is essential to incorporate technologies such as generative AI into the app development process. This study is meaningful in that it identified the types of content preferred by users through a quality evaluation and suggested directions for developing apps that meet current needs. We expect our results to serve as guidelines for creating evidence-based data and developing health promotion apps in the public sector. Further research focusing on the long-term usability, effectiveness, and user engagement of mental health apps—through extended evaluations from a user perspective—could contribute to improving their overall quality and impact.