Since the Apple App Store (AppStore) is the major representative in the market, we first created an app repository based on all the health related apps from the AppStore. The list of apps was crawled from the Apple iTunes Web pages [10], including the pages for the Health & Fitness [11] and the Medical [12] subcategories. Then using our own crawling program, we extracted detailed information of each app via the iTunes Search app program interface (API) [13]. We noticed the results from our data extraction step are in the JavaScript Object Notation (JSON) [14] format. For the convenience of researchers, we transferred the files from JSON format to tab-delimited text files encoded with “utf8mb4” (flat files with array format), so that researchers can directly import these files to Excel or another program for ease of analysis. In the text files, each row corresponds to an app with 39 features, including the app unique identity (ID), app name, description, user rating count, average user rating, etc. Table 1 lists all the 39 features along with their annotations.

Since the Google Play Store (GooglePlay) is now the biggest app store in the market, we also created an app repository based on the information of all popular health-related apps from the GooglePlay. The list of apps was crawled from the GooglePlay Web pages [15], including the pages for the HEALTH_AND_FITNESS [16] and MEDCIAL [17]. We then extracted the detailed information of each popular app using the python HyperText Markup Language parsing tool via the Google Play Search API. For researchers’ convenience, we provided both the JSON format and tab-delimited text files as well. In the text files, each row corresponds to an app with 27 features (Table 2), including the app unique ID, app name, description, user rating count, average user rating, etc. Files in both formats (JSON and tab-delimited) can be obtained from the repository website [18].

In the US market, there are 74,211 apps listed in the Apple iTunes Health & Fitness and Medical subcategories as of December 4, 2014. By removing duplicated entries, we obtained 62,621 totally unique apps in these two subcategories. We note the category of each app is defined by the app’s owner (developer or seller) and approved by Apple’s customer service, so the app categorization was done in the server side (API) and was used directly as our app selection criteria. The primary categories of some apps are neither Health & Fitness nor Medical, but others, such as Lifestyle, Education, Sports, Food & Drink, or Games. To reduce the ambiguity, we only included the 47,883 apps with either Health & Fitness or Medical as their primary category in our app repository. In addition to the US market, this repository contains the information of mHealth apps from the AppStore distributed in four other countries with the most established Internet markets [19]: (1) China (CN), (2) Japan (JP), (3) Brazil (BR), and (4) Russia (RU). There are 27,157 and 21,607 unique apps in the categories of Health & Fitness and Medical from the top five countries of the AppStore, respectively, leading to 48,764 totally unique health-related apps from the top five countries. In both categories, there are slightly more apps available in the United States than in any of the other four countries (Table 3). Overall, more than 98.19% (47,883/48,764) of these unique apps are available in the United States.

The repository also contains information of the most popular apps from the GooglePlay in the United States. For the GooglePlay, the Web pages only list the most popular or the newest released apps in each category based on their release dates and daily user usage. Since the GooglePlay Web pages are updated daily, to get a comprehensive list of all the apps, we collected the app IDs available on the GooglePlay with our crawling program every day from July 24 to December 6, 2014, and combined the results to get a list of 14,817 unique app IDs. We then excluded the inactive apps that are no longer available on the GooglePlay. In addition, as we did for the AppStore, we also excluded the apps with their primary category other than HEALTH_AND_FITNESS or MEDICAL. Finally, we obtained a list of 12,272 totally unique apps, including 6894 and 5378 apps in the subcategories of HEALTH_AND_FITNESS and MEDICAL, respectively. Table 3 gives the total number of apps and the total number of user ratings received in each category. Considering the fact that the GooglePlay apps in our repository are among the most popular ones, and GooglePlay represents the biggest app store now, it is not surprising to see that the number of user ratings received for the Health & Fitness apps in GooglePlay is more than two times higher than the sum of user ratings collected from the top five countries for the AppStore apps in the same category.

According to Table 3, we can deduce that the average number of user ratings received per app in the Health & Fitness category is significantly higher than that in the Medical category, regardless of app stores. Because the number of user ratings reflects the popularity of the app, this comparison result indicates apps in the Health & Fitness category are more popular than those in the Medical category. We further investigated the effect of app prices on their popularity among users. Overall, a majority of mHealth apps are free, especially in the GooglePlay, as high as 74.78% (5155/6894) of apps in the category of Health& Fitness are free apps. Based on Table 3, we can see that the average number of user ratings per free app is always higher than that for a nonfree app. In addition, if we use the average number of user ratings per app as a measure for app popularity, the significantly high percentage of user ratings provided by GooglePlay free app users (95.57%, 11,298,225/11,821,720) suggests the GooglePlay users prefer free apps, compared to the AppStore users.

Based on the release date information of each app included in our repository, we can analyze the trend of mHealth apps available in the AppStore. We plotted the number of apps released in each quarter since the third quarter of 2008 (Figure 1 shows this). From this figure, we can see that the apps in the Health & Fitness category show a quadratic growth (R² = 0.9867), while the apps in the Medical category demonstrate a linear growth (R² = 0.9823). The patterns in the top five countries are similar for both the Health & Fitness and Medical subcategories. The GooglePlay doesn’t contain the released date information of apps; instead, only the updated date information is available. More than 76.57% (9397/12,272) of the apps were updated in the last two quarters. Therefore, the trend for apps in GooglePlay was not analyzed in this study.

The mHealthApps repository allows us to analyze thousands of apps in the market systematically and efficiently, and can be utilized to provide an overview of the trends for mHealth apps. The repository is scheduled to be updated quarterly. Detailed information of all these apps can be freely requested from the repository website [18], but will be restricted for personal and noncommercial use only. A unique feature of our repository is that it provides a new dimension of information of apps, such as the user behavior, which is neglected by many other studies in the field. The user behavior data, including the average user rating, the number of user ratings received per app, and the distribution of user ratings in the five-star rating system are based on millions of mHealth apps users worldwide, and have been tested on the real market. The repository also contains other information, such as the price, the released/updated date, and the app descriptions, which can be used for further business marketing, activity analysis, detail subcategories decomposition, and so on.

It is noted that our study has some limitations. First, the category of each app is submitted by the app’s owner and approved by the app store. Therefore, the accuracy of app categorization is beyond our control. Additional strategy based on nature language processing would be necessary to ensure all the apps included in our repository are health-related. Second, we only retrieved mHealth apps from the two most established system platforms, the iOS (AppStore) and the Android (GooglePlay), there are also apps from other platforms, such as the Windows Phone Store [20] and the BlackBerry World [21]. Third, our repository is limited in the regions the information was extracted from. For the AppStore, we only extracted apps information from the top 5 regions according to the market size, which neglects information from other well developed countries such as Australia and European countries (different stores are separated by different languages), as well as from fast developing regions such as Africa and India. For the Android platform, we only extracted apps information from the GooglePlay US store, due to the complex Android markets in other countries. For example, in China, the major Android stores include Baidu Shouji Zhushou [22], Tencent Yingyongbao [23], and 360 Shouji Zhushou [24], while the GooglePlay is not among the major Android stores. Fourth, the number of apps from the GooglePlay is limited due to the availability of apps on the GooglePlay website, which only lists up to 600 of the most popular apps every day. Our repository is based on the lists of apps accumulated between July 24 and December 6, 2014. In spite of these limitations, we expect this mHealth app repository will not only serve as a centralized information resource for researchers to perform meta-analysis on current apps, but also provide guidance for future research designs in the mHealth field.