Understanding Device Integrations within Diabetes Apps: App Features and Reviews Analysis (Preprint)

BACKGROUND Diabetes management involves a large degree of data collection and self care in order to accurately administer insulin. Several mobile apps are available that allow people to track and record various factors that influence their blood sugar levels. Existing diabetes apps offer features that enable integrations with various devices that streamline diabetes management, such as continuous glucose metres (CGM), insulin pumps or regular activity trackers. While this reduces tracking burden on the users, research highlighted several issues with diabetes apps, including issues with reliability and trustworthiness. As pumps and CGM are safety-critical systems – where issues can result in serious harm or fatalities – it is important to understand what issues and vulnerabilities could be introduced by relying on popular diabetes apps as an interface for interacting with such devices. OBJECTIVE As there is lack of research examining in detail the integrations and potential suitability of apps as part of a wider self-management ecosystem, our goal was twofold. First, we aimed to understand the current landscape of device integrations within diabetes apps and how well they meet users’ needs. Second, we identified the key issues users of the most popular apps face currently and what features are the source of these issues. METHODS Through searches in Android app stores, we systematically identified 21 diabetes apps that offer integrations. We conducted a detailed analysis of 602 user reviews. For each review we recorded its sentiment, features and issues, and additional contextual information provided by the review writers. We used descriptive statistics to analyse the features and issues. We also analysed the reviews thematically to identify additional trends related to the context of use and consequences of issues reported by the users. RESULTS The reviews focused on key features that users found the most important, including: device integrations (43% of reviews), tracking (32.2%), data logging (14.3%) and notifications (11.6%). We found that 54.5% of the reviews were negative (vs 31.1% positive and 14.6% neutral or mixed) and the majority of reviews (62.8%) mentioned issues. The biggest issues related to device integrations included: inability to connect with external devices (25.1%), inability to store, manage or access data (22%), unreliable notifications and alerts (9.2%), issues caused by or related to software updates (8.5%), hardware issues (6.4%) and issues with accessing the app, related services or associated hardware (3.2%). CONCLUSIONS Apps for diabetes management are a useful part of self-care only if they are reliable, trustworthy, reduce burden and increase health benefits. Our results provide a useful overview of desired features for diabetes apps alongside key issues for existing integrations and highlight the future challenges for AP system development.