Table 4.
Usability themes and subthemes.
| Themes | Subthemes | Example Quote |
|---|---|---|
| Ease of understanding | Difficulty interpreting prediction output (6 of 12 participants) | “The prediction score … I don't know how to contextualize that. Is 75% to 100% when we consider screening? Some kind of scale would be helpful.” (Participant 7) |
| Ease of use | Desire for electronic health record integration (7 of 12 participants) | “I don't know if I would necessarily click on a link to go to another website. It is just one step removed that may decrease compliance.” (Participant 5) |
| Reducing clicks while navigating (8 of 12 participants) | “It took me three clicks to get here; it would have been better if it was simpler. It does affect my usage and efficiency and maybe satisfaction.” (Participant 2) | |
| Acceptability | Improving peripheral artery disease diagnosis (9 of 12 participants) | “[I]t would help remind me, especially for more complicated … it can be difficult… when you're just kind of inundated with a lot of different problems in a single 30-min visit … to continue to have this on your radar (Participant 11)” |
| “If you have somebody … on the borderline, … it would be nice to see whether you know your overall gestalt matches with the gestalt of the of the computer… If it was like 83% … I would maybe order that test.” (Participant 9) | ||
| Low priority in diagnosing peripheral artery disease amongst primary care physicians (4 of 12 participants) | If I'm trying to do other screenings, I have to put that risk benefit ratio in the context of everything else. If they haven't had their colonoscopy or mammogram, do I send them for that if they have limited bandwidth? Or do I send them for a PAD screen? (Participant 4) | |
| “[Peripheral artery disease] … it's not like coronary disease, where if you miss it, somebody is going to have an acute event and … death could ensue, right? Whereas if you have peripheral vascular disease that you haven't picked up and they're not symptomatic …, is it really going to make a big difference?” (Participant 1) | ||
| Varying perceptions of machine learning (2 of 12 participants) | “I'm for machine learning to help me be a better doctor. It's going to help me not miss diseases, and it's going to help me manage diseases better.” (Participant 4) | |
| “I think it's hard to know what to make of any particular AI prediction… I …would like to have a link to maybe a paper that was peer reviewed saying … this works” (Participant 11) |