Feature Requests

The ability to provide different instructions for how information provided in the 3 different sources of context are used.
The context section includes 3 different components. 1. The previous sessions in Heidi from the same patient that are attached internally. 2. Documents and files that are uploaded and attached from external sources. 3. The text that is manually entered into the text box on the context tab. It would be very helpful to have a way to provide the AI model with 3 different instructions specific to each of these 3 sources. External notes attached for background are useful for things like past medical history, investigations, prior medical course, social history, family history, and maybe even medications if the context is not too old. It is not at all helpful for things like physical exam, active issues, or impression and plan. Any impression provided in a note I have attached is potentially outdated, incorrect, and certainly does not actually represent my own personal impression. If this impression is to be used in the current session, it should be attributed to a 3rd party (i.e. "Dr. X thought that the patient's symptoms were due to Y") In contrast, I often enter text into the context free text box that provides current context to the present session. For example, even when I describe what I am doing as I perform the physical exam, the voice transcript typically does not capture the full extent and nuances of the physical exam I perform, and this is missed in the note. Instead, I will often type aspects of the physical exam into the context box. I need a way to instruct the model that those aspects of the physical exam are both current and reliable, as opposed to an historical physical exam done by someone else at some time in the past which is buried in a past consult note attached for background. I might also provide Heidi with a list of issues and a very brief impression as a starting point for the model to work with, which reflects my current impression as opposed to some other physician's past impression. Previous sessions added to context should be interpreted differently from the other two sources. This information is historical, but is of known reliability and also reflects my own impression as opposed to someone else's impression. While it is necessary/helpful to instruct the model to disregard historical physical exam findings (as noted above), it would be helpful if the model could reference my own historical sessions (i.e. "edema today is 1+, improved from 3+ last month"). Without the ability to provide different instructions for the different sources of context, this also is not possible.
0
·
Context
Load More