Guiding patients to digitally-interpreted test results with AR

Product Design
Augmented Reality
Prototyping

BD’s COVID test enables patients to scan a test strip using their phone and receive a digitally interpreted result. This technology provides a highly accurate reading, but requires a digital image that meets very particular criteria.

Users were having difficulty following the on-screen guidance while capturing the image, often resulting in a failed test and no result. To improve the scanning success rate, we decided to rethink the on-screen feedback.

Concept Prototype

Challenge

Improve scanning success rates among at-home users.

Strategy

Provide more concise visual feedback to guide users to the optimal test image.

Ambiguous error triggers render unclear user instructions

In early versions of the algorithm, it was unclear exactly which environmental conditions triggered errors - resulting in unhelpful guidance to the user.

I worked with the computer vision team to catalog all possible errors and document their distinct causes. Once we understood what was preventing a successful scan, we could clearly articulate how a user could remedy the problem.

Action-oriented verbiage

In the original scanning interface, users were presented with a description of the technical problem that was occurring. Instead, by using verbiage focused on the action a user could take to resolve the issue, users could react faster with less need to understand the nitty gritty details.

We also reordered the sequence of image quality checks to begin with the most drastic and common problems - thereby ensuring users were in the ballpark before moving on to granular adjustments.

Timely feedback

With only a few minutes to capture an image, it’s crucial that users are only provided with the information they need in that particular moment.

To this end, we created two variations of each error. The first has very simple directions on how to resolve a problem. The second is surfaced once a problem has persisted for more than 10 seconds, and elaborates on what is causing the problem so the user is better prepared to solve it.

It was also crucial that when feedback is presented, it persists long enough for the user to comprehend it before disappearing.

Written instruction < visual guidance

Describing light conditions or an object’s position in 3-dimensional space is incredibly difficult to achieve through words alone.

We took advantage of AR feedback and static graphics to communicate more efficiently with the user. In some cases, animation helped provide direct, concise guidance without overloading the user with information.

Evaluating improvements with users

I prepared a mock prototype of the initial concept for formative evaluation. We then handed off the design to the engineering team, which developed a series of functioning prototypes which we were able to evaluate in a controlled study using contrived test cassettes.

Participants showed an improved scanning success rate, reduced time-to-result, and less confusion with instructions when compared with the original scanning interface.

Outcome

Faster results and a higher success rate

The engineering team put together a functioning prototype of the improved scanning interface which we were able to evaluate using contrived test cassettes. Participants showed an improved scanning success rate, reduced time-to-result, and less confusion with instructions when compared with the original scanning interface.