Supporting Smartphone-Based Image Capture of Rapid Diagnostic Tests in Low-Resource Settings

Devesh Sarda, Chunjong Park, Hung Ngo, Shwetak Patel, Alex Mariakakis
PDF
RDTCheck guides users through the instructions of Quidel’s QuickVue Influenza A+B test and ensures adherence to the procedure using computer vision.

Abstract

Rapid diagnostic tests are point-of-care medical tests that are used by clinicians and community healthcare workers to get quicker results at a better cost compared to traditional diagnostic tests. Distributing rapid diagnostic tests to people outside of the healthcare industry would significantly improve access to diagnostic testing; however, there are concerns that novices may administer rapid diagnostic tests incorrectly and thus be left with invalid results. In response to this concern, we propose RDTCheck — a mobile application that guides users through the instructions of Quidel’s QuickVue Influenza A+B test and ensures adherence to the procedure using computer vision. RDTCheck provides users with real-time feedback so that they may either correct their mistakes or re-administer their test. In this work, we conducted findings from a pilot study that demonstrates how well RDTCheck is able to detect common mistakes and successes during the various steps of the QuickVue test. For the 7 participants we recruited, RDTCheck had an average success rate of 91.1% at giving the correct feedback during the RDT administration procedure.