Table 4 Pharmacists’ willingness to accept AI recommendations
Perception | Theme | Example Quote |
|---|---|---|
Agree | Similar dose scheme | “I could agree with the model recommendation. There really there so close, I don't, although the interval for the model was slightly higher, which, which, you know, sometimes that might be the case.” |
“if it more closely resembled my dosing, if it was just a little bit off, I would be more likely to use it.” | ||
Concessions to AI | “If a model told me that dose would be acceptable, I probably would kind of split the difference.” | |
AI is better | “I would change the dose because I was on the higher side and would want to minimize renal injury. It’s slightly lower than my dose. I prefer the AI’s recommendation and would have chosen that” | |
Disagree | Don't trust the AI | “Oh, I think I would follow my dose. I don't think that trained model recommendation because I've given his age and his body weight and is severity illness, I would want to be a little more aggressive.” |
Lack of dynamic evaluation | “I do not want to change. I always start patients aggressively and then monitor closely. So even if I start q8h, and then I find that creatinine is worse, or hemodynamics look poor, and they had decreased urine output, I would make a decision to check a level sooner. However, in the ICU, you know, initial creatinine is just a starting point; we have to follow up and double-check and triple-check.” | |
Lack of in-depth analysis | “I wouldn't want to change my dose at this point until we got cultures back for the simple fact if we don't know what and sensitivities are going to be. She is a very complicated clinical course, and I do not want to make it worse by underdosing in the beginning.” | |
“This seems to be painful to me, and I would want to have the higher dose. because it's an infection of the knee joint, which means you need a high target range to get a high enough concentration of the drug at the actual site of the infection.” | ||
Not all data can be consolidated | “I wouldn't change it. I will keep it the same. You really need to correlate the numbers you see on the labs with the patient's clinical picture.” | |
Risk index | “So as a result, I think I won't change my dose unless you told me that I had a 95% chance of overdosing. I'd want to know what that's based on.” | |
Vial sizes | “Sometimes we think about vial sizes also, so we don't waste medication and ease of dosing.” |