Diabetes Link: Platform for Self-Control and Monitoring People with Diabetes

To analyze the potential of the proposal and subsequently define its scope, clinical researchers from the CENEXA (UNLP-CONICET-CeAs CICPBA) ¹¹1www.cenexa.org were contacted, due to their long career devoted to diabetes epidemiology, care programmes and derived costs. Their advice allowed us to learn about basic concepts of the disease and multiple aspects of its treatment. In addition, they emphasized the importance of keeping an adequate record of relevant data to achieve a good metabolic control.

Two physicians specialized in DM (with experience in therapeutic education of people with diabetes and co-authors of a manual oriented to the complementation of this activity [7]) were also interviewed. They provided useful information related to the statistical indicators that specialists frequently analyze when assessing the status of a patient and adjusting their treatment.

Finally, this development phase was complemented with a state-of-the-art study of mobile applications oriented to diabetes, which allowed us to discover possible uncovered needs in the area. It is important to remark that this study was repeated once again after the platform was released to carry out the comparative analysis of Section 4.

For the first release of Diabetes Link, a set of functional and non-functional requirements of the platform was agreed among all the participants. Requirements analysis and design modelling were required phases in the plataform development to guarantee the quality of the final product.

The application search and selection process for our comparative analysis was carried out following guidelines proposed in similar studies [1, 3, 5]. The study focused just on the Android operating system due to two reasons: first, Android is currently the most popular option, having more than 70% of the market share [10]; second, even though the mobile application of Diabetes Link will be soon available in iOS, now it is just present in Android.

Today, Google Play does not offer filters to apply to search results. In addition, search terms are matched against app title and/or description, which usually leads to false hits, sometimes caused by spam techniques. To ensure that our analysis is representative, the search on Google Play was carried out on a specific date (Feb. 18, 2020), and all apps found were recorded. The phrase “diabetes control” was used since it is sufficiently general to ensure that every relevant app is detected. As a result, 246 applications were listed (incognito mode on). To be included in the analysis, an application must meet the following characteristics:

• •

  It must offer functionality that facilitates self-monitoring and control by people with diabetes;

• •

  It must allow blood glucose registration as a minimum requirement;

• •

  It must offer Spanish or English language interfaces.

After filtering out those apps that did not meet these requirements, 75 applications remained ¹⁰¹⁰10 Considering the applications that were not used, 2% offer functionality for control, but do not support Spanish or English; 67% are about DM, but only for educational purposes; and 31% are not relevant to this study (poorly classified).. In order to evaluate these applications, the first 10 were selected according to three criteria: (1) greater popularity (order of appearance on results list); (2) higher score; and (3) greater number of downloads. The 3 partial listings were combined to obtain a single final one. In total, 22 applications were reviewed (30% of the total).

After selecting the applications for the comparative analysis, their primary features were surveyed and analyzed. We have considered previous similar studies [1, 3, 5] to determine the feature selection criteria. This comparison allows to quickly and easily see their strengths and weaknesses.

Table 3 presents the main features of each app regarding Google Play information (order of appearence (#), number of downloads, score and version), web access support, its adquisition cost, its communication features, and if it offers diabetes information. First of all, it can be noted that just 32% of the apps offers web access. As described in Section 3, Diabetes Link is cross-platform and both mobile and web accesses are supported. The former becomes more significant at the time of analizying the information and statistical graphics, since its more user-friendly view.

In relation to adquisition costs, 78% of the applications studied require paying a monthly or annual fee to access advanced features (for instance, calculating statistics based on the information entered, adjusting target values, exporting reports or sending them by e-mail, supporting connectivity, among others). As regards connectivity, just four applications (18%) offer this star feature but only one does it for free. Other four applications do not offer any feature for communication support while the rest (14 apps) rely on generating PDF/XLS reports or sending them by e-mail. Diabetes Link, on its behalf, supports connectivity and offers all features for free.

Table 4 summarizes the funcionality features of each application regarding data register and statistical analysis. Each app was surveyed to determine if it supports recording blood glucose, insulin, carbo-hydrates, body weight, blood pressure, medication, and physical activity. Additionally, wether the application analyzes blood glucose, body weight, and blood pressure data or not. Last, if it offers a tabular summary of registered data or not.

Between 59% and 82% of the selected applications allow recording measurements for blood glucose, insulin, carbohydrate intake, physical activity, food and medications intake. In addition,  77% allow recording body weight, and less than 73% allow recording blood pressure. As for the ability of adding labels or notes to the records, all applications offer this feature. Diabetes Link, on the other hand, allows recording of all of these measurements.

All of the applications that were reviewed offer statistical charts/tables for blood glucose measurements. As regards the other parameters, the percentages of applications that analyze them are 59% and 41%, they correspond to body weight and blood pressure, respectively. Furthermore, only 27% are capable of generating summary tables to report all recorded values. In contrast, Diabetes Link generates statistics for all these parameters, as well as a weekly summary tables for blood glucose, insulin and carbohydrate intake measurements, as well as physical activity done. It should be noted that the idea of adding these weekly summaries was suggested by the physicians during their interviews. They stated that they usually ask their patients to provide written records (diary) with this information to analyze patient status and adjust treatment accordingly.

From Tables 3 and 4, it can be observed that none of the applications provides information about diabetes, statistics and connection with monitors simultaneously for free. In that context, Diabetes Link sets itself apart by providing all these features at no charge.

Finally, Table 5 shows the customization features of each application considering user language, target values and units of measurement. First, 22% of the applications does not include multi-language support. Having this feature allows Diabetes Link to offer an expanded reach, which means more people can use it. Both target values and units of measurement can be configured in most applications; in particular, 86% for the former and 82% for the latter. Diabetes Link support these two configuration options allowing users to customize their level of control and preferred units of measurement.

Due to the great number of existing diabetes apps and the large amount of work required to revise them, some limitation of the comparative analysis were imposed. First, the review study just considered the Android OS. Although Android has more than 70% of the market share [10], currently available mobile applications for other OS’s (especially iOS) were not considered.

In addition, the comparative analysis carried out only gathered the presence or absence of a given feature, as a way of simplifying the large collection task. In that sense, neither the quality of functions nor their effectiveness were evaluated. However, useful information could still be extracted.