Related Experiment Videos
Informatics in Radiology (infoRAD): radiology report entry with automatic phrase completion driven by language
1Russell H. Morgan Department of Radiology and Radiological Science, Johns Hopkins University School of Medicine, Baltimore, Md, USA. jeng@jhmi.edu
Summary
This study introduces an automated phrase completion system for radiology reports, significantly reducing keystrokes. The language modeling tool enhances efficiency in medical documentation by predicting and inserting phrases.
Area of Science:
- Medical Informatics
- Natural Language Processing
- Radiology
Background:
- Keyboard entry for radiology reports is still common despite speech recognition advancements.
- Efficient and accurate report generation is crucial in healthcare settings.
Purpose of the Study:
- To develop and evaluate an automated phrase completion system for radiology report entry.
- To assess the impact of language modeling on reducing keystrokes during report generation.
Main Methods:
- A text editor with an automated phrase completion feature based on a trigram language model was developed.
- The system predicts words or phrases based on context and keystrokes.
- Phrase lengths were optimized to minimize expected keystrokes.
Main Results:
- The phrase completion system reduced average keystrokes per report from 194 to 58.
- This represents an average reduction factor of 3.3 (95% CI, 3.2-3.5).
- The algorithm demonstrated a statistically significant reduction in keystrokes (P <.00005).
Conclusions:
- Automated phrase completion significantly streamlines the creation of radiology reports.
- This technology offers a practical solution for improving efficiency in medical documentation.
- The system's performance indicates a substantial improvement over manual keyboard entry.