Credit: Photo by Vitaly Gariev on Unsplash
I’ve been reading more about how AI is being used in clinical care, and a few things caught my attention. In my previous blogs, I talked about preparing for AI and how hospitals are using it behind the scenes. But what about the work healthcare professionals do with patients every day? I see possibilities in clinical reasoning, medication safety, and documentation. I wanted to look beyond what AI might be able to do and see whether there’s evidence that it’s actually helping.
One area I find particularly interesting is clinical reasoning. When evaluating a patient, clinicians use the symptoms, history, examination, and available tests to develop an initial diagnosis. I’ve been thinking about using AI as a devil’s advocate during this process. Imagine asking AI what findings don’t fit your diagnosis or what assumptions need more evidence. A 2026 study involving 70 physicians tested AI-assisted diagnostic reasoning using simulated clinical cases. Physicians using conventional resources scored 75%, compared with 82% when AI was used as a second opinion and 85% when used as a first opinion. That’s a meaningful finding because it shows how AI could support clinical thinking, even though the study didn’t measure real patient outcomes.
Medication safety is another area worth looking at. When prescribing medications, healthcare professionals have to consider possible drug interactions, allergies, and appropriate doses. AI could help review medication orders and identify problems that need attention. A 2025 study published in Cell Reports Medicine compared AI alone, pharmacists alone, and pharmacists working with AI to detect prescribing errors. The combination performed best, with 61% accuracy. That result also shows how much room there is for improvement. Still, I can see the value of having another tool to help review complex medication decisions.
I’ve also noticed changes in documentation during my own healthcare visits. I’ve seen tools being used to record conversations and help prepare clinical notes. Instead of spending as much time typing, clinicians can review a summary of the conversation and make corrections. A 2025 study in JAMA Network Open looked at AI documentation assistants across six healthcare systems. After 30 days, clinicians reported less documentation burden and lower burnout, although the study couldn’t prove AI alone caused those improvements.
These examples show that AI doesn’t have to make the final clinical decision to be useful. It could challenge an initial diagnosis, help check a medication order, or prepare a clinical note for review. These may seem small parts of clinical work, but they take time and attention every day. Of course, these tools still need to be accurate, carefully evaluated, and used with proper clinical oversight. I think the bigger question is how we can bring these benefits into everyday clinical practice without adding more work for healthcare professionals.
If you work in healthcare, which of these uses would make the biggest difference in your daily work? I’d love to hear your thoughts in the comments.
Leave a Reply