The number of parties involved in the consultation:
Brevi is currently only able to support two participants per encounter. For best results, please limit sessions to only 2 main speakers.
Software does not work with most pediatric and physical therapy cases and performs poorly in complex psychiatric encounters and complex emergency cases. Software can not come up with text or scribe that the user(s) didn't speak/mention during an examination.
3rd Party Description Limitations:
Do not use Brevi if a 3rd party talks on behalf of a patient. Brevi’s Natural language Generation model disregards sections with 3rd party frame of reference:
- e.g., when a physician or patient uses 3rd party pronouns, the software does not capture those parts of the dialogue and therefore does not include them in the summary.
- The Brevi NLG Model is designed and tailored to keep the naturalness of Physician-Patient Clinical Dialogues as they used to and do not follow any rules. To implement this technique, the model excludes parts of conversations unrelated to health concerns and discussions about third parties from dialogues that resulted in the abovementioned limitations.
Pronouncing Drug Names:
In many instances, patients and providers have trouble pronouncing some drug names. This may cause speech-to-text transcription errors, consequently causing the same mistakes in summaries.
e.g., - "Molnupiravir" may be pronounced wrong by most providers.
Conclusion: Please check summaries more carefully if the visit discusses medications with complicated names.