This project will investigate improvements to models for automatically generating physician notes from transcripts of conversations between physicians and patients. The model will capture important medical information such as symptoms, treatments, and medications, and automatically formulate a free-text note that mimics the style of the physician’s manually entered notes. The model will use context and time information to ensure that the documentation is accurate and complete.
With the rise in electronic health records, we are seeing more reports of clinician job dissatisfaction that is attributed to the amount of data entry work they are now required to do. Therefore, there is a lot of interest to use a speech recognition system to reduce the burden clinicians have by automating the documentation process as it passively listens during clinical visits. However, while many companies and research groups are devoted to construct such a system, there has not been any evaluation on the practicality and usability of this approach.
Join a thriving innovation ecosystem. Subscribe now