Language Use in Clinical Phrase Prediction Systems
Author Information
Author(s): Zaghir Jamil, Bjelogrlic Mina, Goldman Jean-Philippe, Ehrsam Julien, Gaudet-Blavignac Christophe, Lovis Christian
Primary Institution: Division of Medical Information Sciences, Geneva University Hospitals, Geneva, Switzerland
Hypothesis
Clinicians adapt their language interactions with a phrase prediction system according to Zipf’s least effort principle.
Conclusion
The study shows that clinicians prioritize communication accuracy over efficiency when interacting with a phrase prediction tool.
Supporting Evidence
- Users adapt their language to minimize effort while maximizing communication accuracy.
- Clinicians' interactions with the phrase prediction tool showed a trend towards more distinctive queries.
- User-label seniority correlated with improved query accuracy over time.
- Medical idioms were initially used frequently but declined as users became more experienced.
- Users demonstrated a preference for accuracy over conciseness in their queries.
Takeaway
Doctors change how they talk to a computer to make sure their messages are clear, even if it takes a little more time.
Methodology
A large-scale observational study analyzing clinician interactions with a phrase prediction tool in a clinical setting.
Potential Biases
Potential biases include technology obligation bias, speed bias, and correctness bias due to the clinical setting.
Limitations
The study's findings may not be generalizable due to biases inherent in the clinical context, such as technology obligation and correctness biases.
Participant Demographics
1,763 healthcare professionals using French language and medical jargon.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website