Introduction
Ambient clinical documentation systems use artificial intelligence (AI) to automatically generate medical notes from patient-clinician conversations, potentially reducing administrative burden and allowing clinicians to focus more directly on patient care. While these systems have become increasingly sophisticated, their accuracy depends on the clarity and quality of the source material.
This guide outlines the challenges clinicians are facing today with regard to documentation systems and the best practices for verbal communication that can optimize documentation output quality.
Today’s Challenges
Ambient clinical documentation systems can extract structure and meaning from long, unstructured conversations with surprising accuracy. They often catch details the clinician did not consciously note but which were in fact spoken.
However, when a conversation is unclear, even to a human listening back, no system can reliably reconstruct the intended meaning. This limitation is the same for humans and machines: clear points, clear decisions, and clear conclusions produce clear documentation. Vague, mumbled, or contradictory speech forces guesswork.
Clinical work requires precision, and tolerance for errors is low. One practical barrier to adopting ambient scribes is the time clinicians spend correcting omissions or hallucinations in the draft notes. While the underlying technology continues to improve, it is unrealistic to expect perfect notes from unclear source material.
Recommended Best Practices
Audio conditions
Basic audio conditions still matter. Much of the model’s performance depends on a clean signal: minimal background noise, limited side conversations, and avoiding long stretches of overlapping speech. Speaking at a steady, clear pace helps preserve the structure of the conversation for downstream interpretation.
Ambient dictation
If the above conditions are difficult to obtain during the consultation, simply dictating a detailed summary after the consultation often produces very good results and allows better control of audio. A nice side effect is that this dictation does not need to be structured like standard dictation, but can follow your train of thoughts as you reflect on the case.
Communication practices
The communication behaviors that lead to accurate ambient documentation align with the commonly used and taught Calgary-Cambridge model for structuring medical interviews.[1] These practices give the ambient system the material it needs to resolve ambiguities.
The key points that support ambient documentation:
Active listening helps the conversation unfold with clarity. This can include short restatements of key points that invite elaboration or correction on misunderstandings. These moments also give the model stable anchor points in the conversation.
Summaries throughout the encounter verify mutual understanding and show the patient that their information has been absorbed. They also resolve small contradictions before they accumulate.
Clear articulation of assessments and plans, including expected outcomes and side effects, is essential for safe care and informed consent. Stating these conclusions in complete phrases rather than trailing comments reduces ambiguity.
Physical examination findings must be stated clearly if they are to be captured by an ambient system. Simple, explicit phrasing works well, e.g. “Heart sounds normal, no murmurs,” or “Left lower lobe with reduced air entry.” If speaking the findings during the examination is inappropriate, summarizing findings afterward also provides the system with the needed material.
Examples
Conflicting symptom severity is a challenge
A patient might first say, “It hurts a little when I breathe,” and later describe the pain as “unbearable.” Without a clarification, the system is forced to choose between mutually incompatible claims.
A simple restatement often clears it up for both clinician and model, e.g. “You mentioned mild pain at first, but later called it unbearable. Why is this? Does it fluctuate?
Medication adherence often shows similar contradictions.
A patient may say, “I take all my pills,” but later admit, “Some days I forget the morning dose.”
Clarifying that the intended regimen is daily but the actual pattern is partial adherence resolves the conflict and prevents inaccurate documentation.
Timelines frequently drift in unstructured conversations.
A patient might start by saying the chest tightness began last night and later recall that the symptoms have been present for one to two weeks
Restating the timeline gives the model a clean, unified interpretation, e.g. “So, it began one to two weeks ago and worsened last night?”
Numerical inconsistencies also arise.
A patient who reports losing “about five kilos” and later says “more like ten” leaves the model with two incompatible numbers unless the clinician confirms which figure is closer to the patient’s true estimate.
Physical examination findings need to be explicit
When speaking aloud is inappropriate or disruptive, the clinician can articulate the findings once the examination is complete, for example while the patient is getting dressed or after they have stepped out.
The key is that the findings are spoken in a clear, explicit way at some point, so the system has an accurate record to work from.
The recording can be stopped and resumed as many times as desired.
Role confusion can significantly distort output.
Consider the situation where a patient says, “I didn’t eat anything yesterday,” and a nurse notes, “The patient met about 40 percent of calorie needs yesterday.” Without signaling who is summarizing and who is reporting personal experience, the system sees two conflicting facts of equal weight.
A simple clarification resolves the tension, e.g. “You feel like you ate nothing yesterday, but our records show 40 percent of your needs.”
References
Kurtz, Suzanne M, and Jonathan D Silverman. “The Calgary-Cambridge Referenced Observation Guides: An Aid to Defining the Curriculum and Organizing the Teaching in Communication Training Programmes.” Medical Education 30, no. 2 (1996): 83–89.https://doi.org/10.1111/j.1365-2923.1996.tb00724.x.
Conclusion
The practices that support accurate documentation are the same fundamental communication skills utilized across clinical care settings. Clear speech, active listening, timely clarification of contradictions, and explicit articulation of findings provide ambient systems with the material and context needed to generate reliable documentation. Clinicians who adopt these communication habits will leverage ambient documentation effectively while maintaining the precision that clinical work demands.
Have a question for our team?
Click Support in the bottom-left corner of the console to submit a ticket or reach out via email at [email protected] and we'll be happy to assist you.