Problem Space

Rapid diagnostic tests (RDTs) are widely used in many low-and-middle-income countries (LMICs) by lab technicians and community health workers for many diseases. However, there is still much to be done to improve the accuracy and efficiency of result interpretation. It is common for operators to have variable interpretations of weak test lines. Additionally, RDT generated data is used by national and international health organizations to make policy and resource allocation decisions. Those decisions rely on accurate and timely reporting of RDT data.

Problem Statement

How might we improve RDT interpretation consistency and efficiency for laboratory technicians and, eventually, community health workers in LMICs using AI and ubiquitous devices?

Secondary Research

Since all team members were new to the healthcare space, especially to diagnostics, I led us in conducting secondary research in the form of literature reviews and competitive analysis. We researched 70 published literatures and websites, and analyzed 6 existing products. Below are some of the common themes we’ve identified:

Personas & User Journey Map

Through user research, we learned that we potentially have 2 groups of target users: lab technicians and community health workers. They all perform both single and batch tests, and our application should allow both workflows. The two groups have slightly different needs due to different educational background and use environment, and those differences are illustrated in the personas below.

Information Architecture

With the personas and journey map as our guide, I led the team develop a product requirement document with prioritization and justification for each feature. From there, I created the IA for our mobile app, shown below.

Paper Wireframes & Mid-Fi

I started the design process by making low-fidelity sketches on paper. Sketches are fast, cheap, open-ended, flexible, and very iterative. With them, I was able to visually present my design thoughts and progress to my engineering and PM teammates and quickly iterate based on their feedback.

Final Design

(in progress)

As of November 2020, I have designed 3 iterations based on pre-alpha and alpha testing feedback. The latest iteration features a Templates → Current Tests → History flow and is meant to improve data collection efficiency for lab technicians. The data collected is currently being used to train our own machine learning model that would eventually detect read window lines, making it an impactful tool for low and middle income countries (LMICs). All designs have been handed off to dev and have been fully developed with Android and Flutter.

(clickable prototype coming soon)

Usability Testing

(in progress)

Currently, I am leading the usability testing efforts of the team. As of November 2020, we have conducted 11 in-person usability studies with 6 RDT experts and 2 non-experts. All tests were done at the PATH office in Seattle and all participants wore gloves in order to best mimic an actual use scenario. We are planning to evaluate our final prototype in early December before launching our application. Below are some of the key insights we’ve identified so far. We also organized everything onto a map to visually prioritize future tasks.

view full usability study report

This case study is a work in progress. Please feel free to contact me for more information.

This was my first time leading the end-to-end UX process as the sole designer on a team. I am responsible for user research, UI design, usability testing, and every single step that is entailed in those stages of the process. Needless to say, this experience has been as challenging as it has been rewarding. Because of the nature of an industry-sponsored capstone project, I learned to work with business and technical constraints. I especially became more cognizant of how “blue sky thinking” may be good as a brainstorming strategy but is not the most appropriate when there are deadlines to meet and other stakeholders involved. I’d also like to thank all of my amazing teammates who have been working hard on the machine learning and app development fronts. Our team works in an Agile environment, and so far we have been meeting all of our sprint goals while rapidly iterating. I cannot wait for our final showcase in December 2020 ✨