Detailed HIV Self-Testing Patterns Derived from Paradata in the mLab App Clinical Trial

Self-testing is a critical component of public health initiatives aimed at slowing and stopping the spread of HIV. It has the promise of accessibility, reliability, and convenience, and because of the benefits derived from its inherent privacy, self-testing may overcome the barriers associated with HIV screening with at-risk populations. Still, questions remain about whether and how self-testing can adequately link patients to care and engage them with other interventions when needed. This presents an opportunity for digital platforms to bridge the gap, connecting patients with the HIV continuum of care. During a recent clinical trial of the mLab App, a mobile health intervention designed to increase HIV testing rates, we collected screen-level paradata-detailed logs of user interactions within the application-focusing specifically on user behavior during the test-result interpretation workflow. Among enrolled participants, 330 HIV self-tests were completed in the app, with 74.2% occurring within the scheduled testing window. Three post-timer screens (Preview Test, Upload Picture, and Visual Result) accounted for 60.6% of incomplete testing sessions, highlighting friction points in the result interpretation workflow. Users who experienced discordant automated results (i.e., when the app's automated interpretation differed from the user's visual inspection) demonstrated reduced subsequent engagement but did not significantly alter future test-taking behavior. These findings identify critical moments in the HIV self-testing workflow and provide actionable insights for improving the design of digital tools that support accurate testing and linkage to care.