Ambient AI in Rheumatology Practice: Less Typing or Less Human? Save

Rheumatology has become a sedentary, computer-based job. We sit at our computers to pre-chart, type during the patient visit, and bring notes home to complete on laptops while sitting on the couch. Do you spend more time looking at a patient or more time looking at a screen? As artificial intelligence increasingly enters nearly every aspect of our society, what will this technology bring to the rheumatology office? Will it help make us better doctors… or simply help reduce our screen time?

 

Ambient AI documentation is being adopted as a tool for clinical documentation, with the potential to improve efficiency and patient interaction. This may be particularly well suited to rheumatology, where visits often involve complex, longitudinal histories, multiple high-risk medications, and extensive counseling. Translating this information into a comprehensive note is challenging while simultaneously maintaining eye contact, recognizing non-verbal cues and responding to patient concerns and emotions.

 

Ambient AI also introduces significant risks. Gerke et al, writing in the AMA Journal of Ethics, highlight concerns regarding informed consent, power asymmetry and equity. Patients may understandably question who is listening, where their conversations are being stored, and whether that information could be used in the training of additional AI models. Accuracy is another concern. Misheard medications, doses, physical findings or other clinical details can alter the intended meaning of the documentation, while AI-generated hallucinations can introduce details that were never discussed. Patient accents, non-English languages, background noise or multiple speakers may further affect transcription accuracy. Legal and regulatory questions exist regarding ownership, storage, retention and secondary use of recordings of patient encounters. A recent case in Washington’s Clark County Superior Court found that patients do not have a legal right to their ambient recorded patient visits, but national regulations remain unsettled.

 

When Ambient AI was recently unveiled at my institution, I was eager to adopt a new technology to ease my documentation burden. My initial experience was frustrating. Early transcripts were poor and led to documentation errors. Adding a high-quality microphone to the room, however, dramatically improved performance, and I found transcription accuracy was exceptional when audio quality was adequate.

 

I also found that the AI-generated assessment and plan useful, but not nearly as useful for clinicians as my own meticulous documentation. I began modifying my speech during visits, deliberately restating ambiguous findings or describing visual findings (ie “you have a rash on the extensor part of the right elbow, but not left”) and explicitly stating medication doses and regimens (ie “you are taking methotrexate 2.5 mg, 8 tablets weekly with a split dosing of 4 in the morning and 4 in the evening”). These changes felt a bit awkward, but the clarifications improved the accuracy and usefulness of the resulting note.

 

To my surprise, my overall documentation time was not dramatically reduced. I still need to review and edit the AI-generated “rough draft” after each visit. Despite these real-world drawbacks, I have found this technology to be a game changer.

 

The more important question is therefore not whether AI makes documentation more efficient, but whether it allows us to be more present in the examination room. I am no longer turned toward the computer, anxiously trying to record every detail. Instead, my body is facing the patient, and I can actively listen to their stories, complaints, and concerns. Ambient AI should not replace clinical reasoning, careful documentation, or the physician-patient relationship. Nor should it be seen as a shortcut to eliminate the physician’s responsibility for the medical record. Every AI-generated note still requires physician review, editing, and ultimately accountability.

 

The promise of ambient AI, at least in my clinical experience, is not necessarily less documentation time. It is the promise of less distraction. The technology has not dramatically changed my efficiency, but it has changed where I direct my attention. If ambient AI can help us spend less time looking at screens and more time looking at our patients, it may counterintuitively increase our humanity as physicians.