At minimum, 120 (78.9%) of the 152 apps assessed did not disclose any expert credibility or use of evidence, such as providing references, citing information sources used, or disclosing that in the process of the app’s development, there was consultation with at least 1 expert, which includes people with lived experience and caregivers. Table 1 illustrates the number of apps that have disclosed these elements.

Out of the 152 apps, most apps (n=114, 75%) were created by miscellaneous app developers, in other words, developers who were not stated to be associated with any organization or institution. Researchers and health care organizations made up the next 2 largest categories of app creators, observed in 12 (7.9%) and 11 (7.2%) apps, respectively. Members of the public created 8 (5.3%) apps, while medical app developers created 5 (3.3%) apps. Nonprofit non-health care organizations created 2 (1.3%) apps.

Of the 152 apps, approximately one-third (n=53, 34.9%) were categorized under “Health & Fitness” by app stores. Following this, “Medical” and “Educational” comprised the next 2 largest categories at 40 (26.3%) and 18 (11.8%) apps, respectively. Multimedia Appendix 3 presents the wide range of categories assigned by 4 app stores and the number of apps tagged under each category.

Table 2 illustrates the wide range of intended users of the assessed apps. Note that some apps were observed to have targeted >1 group of users. Of the 152 apps, 105 (69%) apps were intended for people with dementia, making them the most frequently targeted audience. Informal caregivers at 79 (52%) apps and the general public (ie, people without neurocognitive symptoms) at 65 (42.8%) apps comprised the next 2 largest groups.

Of the common mHealth app purposes, the most common was providing education on the health topic (ie, dementia), applicable among 53 (34.9%) out of the total 152 apps assessed. As illustrated in Table 1, the least common was providing professional supervision to users, applicable among 4 (2.6%) apps.

Of the lifestyle elements important for the prevention of dementia, the most common category was mental stimulation, addressed by 59 (38.8%) out of the total 152 apps assessed. The least common element was sleep, addressed by 3 (2%) apps. Table 1 illustrates these findings. Meanwhile, 65 (42.8%) apps did not focus on any of the lifestyle elements. In addition, only 6 (3.9%) apps addressed sensory health. Table 3 also presents the number of modifiable lifestyle behaviors addressed by the included mHealth apps.

Observed in 57 (37.5%) out of the 152 assessed apps, providing communication avenues with various parties was the most frequently observed app feature, while 45 (29.6%) apps incorporated game-based elements. The least embraced feature, found in only 2 (1.3%) apps, was the ability to link to social media.

In addition, there was a lack of mention of features designed for people with dementia such as visuals supplementing texts, voice recognition, and customized accessibility features in the app descriptions.

Of the 152 apps, 136 (89.5%) apps specified their creation date or year and the date of their last update. At the time of analysis in April 2023, only 14 (9.2%) were modified within the previous month.

Most apps (143/152, 94.1%) had disclosed app ownership, as seen in Table 1, and most apps (132/152, 86.8%) did not disclose any author affiliations or sponsorships or credited any authors or lack thereof.

The most disclosed facet of user privacy and security was whether user data were being collected. This was disclosed by 79 (52%) out of 152 apps, of which 45 (57%) apps collected user data. The vast majority of apps (131/152, 86.2%), however, did not disclose if any user data were encrypted at all, with an average median of 105 (69.2%) apps that did not disclose the privacy and security of their apps. Figure 2 visualizes these observations.

Of the 152 apps, most apps (n=111, 73%) were free. Free with in-app purchases was the second largest category at 25 (16.4%) apps, followed by 13 (8.6%) premium or paid apps. The smallest category stood with 3 (2%) “freemium” apps, referring to those with basic or limited features available for free download but offering a paid subscription for premium features once installed.

Many apps were available on >1 app store. Figure 3 presents the number of apps available by store, which include the Google Play store, App Store, Microsoft app store, Samsung Galaxy store, and combinations of these.

Due to the structure of some app stores (ie, certain elements of app information are or are not made available and provided by app stores), 120 (78.9%) out of the 152 assessed apps had no ratings available. The next highest group of app ratings was 4 to 4.9 stars, given to 14 (9.2%) apps. Only 7 (4.6%) assessed apps received a rating of 5 stars.

A total of 77 (50.7%) apps had no download information available due to the structure of some app stores (eg, the App Store did not present this element of app information at the time of charting), while 20 (13.2%) had between 100 and 999 downloads. The next highest number of downloads were those apps with <100 downloads as observed in 19 (12.5%) apps, then between 1000 and 4999 downloads observed in 18 (11.8%) apps. Only 4 (2.6%) apps had between 5000 and 9999 downloads, while 10 (6.6%) apps had between 10,000 and 49,999 downloads. Another 4 (2.6%) apps had between 50,000 and 999,999 downloads.

As the Silberg scale aims to assess the accountability of web-based health information, with each point representing 1 of 9 elements of accountability disclosed [29,30], a low score suggests that these apps possess gaps in accountability to users. More than half of the apps (84/152, 55.3%) evaluated scored 2 out of 9 Silberg scale points (mean score 2.4, SD 1.0 points), as visualized in Table 4.

Of the 152 apps, 4 (2.6%) apps did not score any Silberg scale points (ie, received total scores of 0), and none of the included apps achieved a perfect score. The Care4Dementia app totaled to 7 points, forgoing 2 elements of app accountability: crediting the app’s authors and indicating whether an update had been made within the last month (as of the screening and review stages). Meanwhile, the Memory Lane Games app received a score of 6, forgoing the same elements as the Care4Dementia app along with failing to disclose author credentials.

Most apps did not report author credentials, give information on whether peer-reviewed papers or other sources were used, or provide references. When creating apps that address people with dementia, there should be clear guidelines for developers to follow to ensure that a minimum standard is met in terms of how credible the information in the app is, such as including the source of evidence-based information presented [31]. The absence of these factors could impact the overall effectiveness of the app and its uptake by patients or could mislead and misinform individuals looking to use apps to improve their brain health.

Similarly, most assessed apps (114/152, 75%) did not specify the app developer association, meaning their affiliations were not clearly defined. Meanwhile, only a small subset of assessed apps (23/152, 15.1%) had been created by developers categorized as researchers or health care organizations. This reinforces the need for more credibility and reporting of evidence-based information that could benefit users. Furthermore, only a few indicated having consultations with an expert, which included people with dementia or caretakers of people with dementia, whose lived experiences could be strongly leveraged to take input from, as apps are developed to influence behaviors and habits [32]. The lack of consultation with experts [33] or people with lived experience could impact the quality of the app and the information delivered, making these apps less relevant to those who could best benefit from it. These gaps may mislead consumers in the target populations or provide them with misinformation that could further affect their condition onset or prognosis.

Of the 152 apps assessed, 96 (62.5%) were categorized under “Health & Fitness,” “Medical,” and “Educational,” with most apps intended for persons with neurocognitive symptoms, informal caregivers, and the general public (ie, people without neurocognitive symptoms). Meanwhile, only a small subset of these apps actually provides information pertaining to dementia; offers diagnostics tools; and provides reminder, goal-setting, progress-tracking, or habit-tracking features. The lack of diagnostic tools and ability for professionals to remotely monitor users align with the minimal evidence-based information or clinical credibility available in these apps. There appears to be a gap in what these apps that claim to target persons with dementia (and more broadly, people with neurocognitive symptoms) realistically provide to benefit this population. Education, reminder features, goal setting, and tracking would be beneficial tools to provide through apps not only for the target population but also for informal and formal caregivers as well (ie, apps could help set reminders for medications while providing education on certain habits that individuals can commit to improving their condition) [34].

Apps were screened for whether they focused on one of the modifiable lifestyle behaviors associated with reducing the risk of developing dementia, which include exercise, sleep, stress management, mental stimulation, eating and nutrition, and social stimulation. More than 2 in 5 apps did not address any lifestyle behavior, while 57.2% (87/152) addressed at least 1. When apps did address modifiable lifestyle behaviors, it was most common to focus on 1 behavior at a time while few addressed a combination of 2 behaviors per app, as demonstrated in Table 1. The lifestyle behavior of focus was mental stimulation; a greater focus on mental stimulation in these apps is in line with the number of apps that had contained game-based elements. This complements the observations of many studies that dementia prevention mHealth apps focused on mental stimulation in maintaining cognitive vitality in healthy individuals and those with mildly impaired cognition, including improved memory and enhanced quality of life [35].

Research suggests that activities addressing these lifestyle behaviors could in fact improve brain health and lower the chances of developing dementia. In a study exploring physical exercise, diet, smoking, alcohol consumption, cognitive stimulation, and social stimulation, Jia et al [36] observed that positively maintaining any one of these lifestyle behaviors contributes to slowing memory decline. Mamalaki et al [37] also observed the individual impacts of diet, physical activity, sleep, and engagement in activities of daily living (including social stimulation) in maintaining cognitive vitality and reducing the risk of developing dementia. Both studies highlighted the combined effects of addressing multiple lifestyle behaviors to further slow cognitive decline, with the upkeep of more behaviors resulting in stronger prevention [36,37]. It is noteworthy to mention that both studies found diet as the modifiable lifestyle behavior with the largest associated effect on maintaining cognitive vitality and reducing the risk of decline. Thus, if apps are to target patients and their caregivers, it would be most useful to use this evidence-based information to allow for habit development and tracking in activities related to exercise, reminders to maintain sensory well-being, or prompts that could structure stress management. In particular, delivering interventions related to the prevention of dementia using the mHealth app format can expand the availability of these products to hard-to-reach populations [38].

Apps were also categorized based on whether they offered game-based elements, communication avenues, and links to social media. These features are attractive and can address feelings of loneliness for patients. While game-based apps can be a therapeutic experience, this population may have difficulty in carrying out the tasks or activities required by some of these game apps. Apps may also be a promising way to address social isolation and loneliness [39], providing a channel for in-app users to connect among themselves or presenting features to link with family, friends [40], or the general public for connection or support.

One subset of the intended users of these apps, that is, persons with dementia, requires specific design features for these apps to be beneficial and accessible. Such features include prompts (eg, redirection prompts while completing an activity), visuals supplementing the text, reminder systems, and voice recognition, along with the option to customize accessibility features because dementia affects each of those with the condition differently [41,42]. If these game-based apps are targeted at patients, they may be inaccessible for them due to impaired cognitive abilities where they cannot complete multistep cognitive tasks, remember information, or answer questions in a short time span as they may be cognitively demanding. These features are clearly lacking, making them less accessible to the target users. Thus, the gap remains that consulting the target populations during the development of these apps to identify useful elements would likely serve as the best use of time and resources to yield a positive impact on patients.

Information on currency, year of creation, and last update date were available for most of the apps. This practice of transparency around app currency is important as new research is released regularly and thus would be important for the consumer to understand when the app itself was created along with the date of the information that the app is based on.

A minority of apps credited authors to acknowledge their work, reported author affiliation to demonstrate connection to an organization or academic center, or disclosed sponsorship on whether a specific organization provided funding for the development of the app. This results in poor transparency, as it does not provide consumers with the necessary information to identify the basis of the app or disclose whether there is a possible conflict of interest because of a sponsorship, author, or organization affiliations. Moreover, it is crucial to disclose all these pieces of information, as simply reporting sources or authors while failing to report sponsoring organizations obscures acknowledgment of possible bias embedded in the information presented and the intention behind mHealth apps [43,44]. Providing this information to users could add credibility to the app, which could help an individual decide whether or not such an app will serve their needs and whether it is trustworthy and safe to use [23,45]. Furthermore, the availability of this information is crucial to ensure patient safety and suitability on part of professionals who may recommend such mHealth apps for clinical use [23,45]. By contrast, users are less aware of and less preoccupied with considering an app’s evidence base and instead emphasize the quality of its consumer-oriented, practical use such as data use, battery drainage, loading speed, and presence of advertisements when deciding whether to continue its use [43].

In regard to privacy and security of these apps, most apps did not provide information regarding the collection of user data, data sharing with third parties, user data encryption, or data linking to users. This demonstrates a gap in the security of app-user data, which may be at risk of being compromised due to weak data governance guidelines that are not disclosed clearly or at all. Close attention should be paid to ensuring the privacy and security of mHealth apps, especially when dealing with susceptible target groups such as people living with mild cognitive impairment [46]. One relevant guideline that can be followed is the mHealth Data Security, Privacy, and Confidentiality: Guidelines for Program Implementers and Policy Makers, which details best practices and considerations for mHealth app development. These best practices cover the data life cycle stages of data capture and storage, access, transfer, and disposal, as well as address risks associated with operating systems, mobile devices, networks, and mHealth data storage [47]. The Privacy and Code of Conduct on Mobile Health Apps developed by the European Commission [48] lays out principles for mHealth app development and function, including considerations for user consent, app security measures, and use of user data for secondary purposes such as research and third-party sharing among others. While these guidelines are not official law or policy, they provide detailed considerations that will assist mHealth app developers in creating and sustaining responsible data governance.

For apps to be useful, they should be accessible to the targeted population. Commercialization of the apps significantly impacts this, as almost three-quarters of assessed apps were offered for free, while about one-tenth involved a purchase for the basic or the full version of the app. Furthermore, 44.1% (67/152) of the apps were offered both on the App Store and Google Play app store, whereas an exclusive 33.5% (51/152) were offered only on Google Play as opposed to 17.8% (27/152) of apps that were exclusively available on the App Store. Interestingly, only a small number of the apps (7/152, 4.6%) were available on the Microsoft app store, and none were offered on the Samsung Galaxy app store. A needs-based analysis to identify which app features are most useful for patients and caregivers would be appropriately supplemented by identifying the type of smartphones that are accessible to this population.

Measured by the number of downloads, user reception is important when it comes to measuring meaningful use of apps. While not all apps had information pertaining to their exact number of downloads, 39 (25.7%) out of the 152 assessed apps had <999 downloads, of which 19 (49%) had <100 downloads per app. This illustrates a very low use or download rate of apps that are categorized and are targeted for persons with neurocognitive symptoms.

Apps were also scored using the Silberg scale, which is an established validated measure to capture the extent of accountability of health information presented on the web, including through mHealth apps [29,30]. An app can score up to a total of 9 Silberg scale points, wherein the higher the total Silberg score, the more the accountability of the information presented by or used to create the app is communicated [49,50]. Meanwhile, a low total Silberg score implies the opposite, suggesting that the app in question does not transparently disclose the credibility of the app’s information, which may therefore lack accountability toward users and the public. Within this study, the mean Silberg scale score of 2.4 (SD 1.0) across all reviewed apps points toward a trend skewed toward low accountability of information used and presented by the apps (ie, in communicating the authorship, attribution, disclosure, and currency of their information). The observed skewed trend of apps with low Silberg scale scores raises concern for a lack of accountability these apps take in disclosing the credibility of their foundations, their reliability, usefulness to targeted users, and the efficacy and effectiveness of their results, if any.

While the apps assessed in this study produced low Silberg scale scores, this does not appear to be unique in this regard, as other nondementia mHealth apps have observed similar findings. Several studies assessing mHealth apps using the Silberg scale found that most apps scored <5 points [29,51-53]. In the case of a study assessing the (then) current state of mHealth apps that addressed postnatal depression, the authors observed a mean Silberg score of 3.0 points and suggested the need for the involvement of health care professionals in developing mHealth apps [51]. In another study, an analysis of obesity management mHealth apps observed a mean score of 4.55 points along with a low percentage of apps disclosing information sources and references [29]. The assessed obesity management apps were found to present poor accountability of the information, highlighting the need to improve the current state of mHealth app information quality. The systematic review of mHealth apps by Feldman et al [53] that had addressed peripartum mood disorders reached a similar conclusion, finding <30% of the assessed apps indicated the use of research for app development. Overall, the authors found that most apps reviewed in their study were of poor quality, as indicated by low Silberg scores, and suggested their poor accountability of information as a hindrance to users in making an informed decision about their quality [53]. Overall, there is a general trend of low Silberg scale scores observed across mHealth apps, which is a gap that requires addressing by developers.

The observations noted in this section point toward the need for app developers and app stores to fill in the gaps when presenting descriptions of the products they are offering. The most significant gaps observed in targeted apps included (1) a lack of prevention-oriented apps addressing >1 modifiable lifestyle behavior (diet, sleep, exercise, stress management, mental stimulation, social stimulation, and sensory health) associated with dementia risk; (2) an absence of apps created using evidence-based information and credible authors and an absence of transparent disclosure of these sources; (3) inadequate rigor of general mHealth data privacy and security procedures and a complete lack of disclosed specific guidelines created for people with cognitive impairment; and (4) a lack of evidence-based, usable app features designed for the needs of people with declining or limited cognitive abilities.