Note
Access to this page requires authorization. You can try signing in or changing directories.
Access to this page requires authorization. You can try changing directories.
As the partner integrating Microsoft Dragon Copilot functionality in your EHR with your own UI, or embedding the Dragon Copilot web UI in your EHR, you can use what we provide or you can customize your user experience (UX) implementation to some extent. For example, some EHRs customize how content is delivered or how AI outputs are presented on mobile and desktop.
To make sure the user experience is well-supported, we recommend the following best practices:
Support multi-device workflows - enable users to work flexibly where they are, on the device that works for them. Don't force a specific workflow.
Support different types of voice interaction - ambient recording, dictation, and voice commands. To provide dictation directly at the cursor, embed Dragon Medical SpeechKit or employ the Dragon Copilot desktop app, which replaces Dragon Medical One. When users have access to Dragon Copilot's ambient recording capabilities, their need for dictation lowers, but is still present.
Automate the delivery of the AI results and syncronization of the results with your EHR; don't add a step for users to indicate they've finished recording.
Implement feature parity across devices - for example, don't limit some features/content to appear only on mobile apps or only on desktop apps.
Make Dragon Copilot easy to access. If you're embedding our web UI, provide the launch button in an easy to access location. Don't, for example, bury it deep within the menu options. If you're building a custom integration, there should be a clear pathway to support the full workflow. The user shouldn't have to navigate deep into the UI to access the full suite of features.
Make sure Dragon Copilot is always available. Authenticate the user as soon as they open the EHR on any device.
If you're building a custom integration, make sure clinicians check they have patient consent before they start ambient recording. The consent status should ideally be tracked by the EHR and displayed in the schedule. If you're embedding our Dragon Copilot web app, our UI includes a consent reminder.
Make it easy to start ambient recording. There shouldn't be a delay in the start of recording. Make it clear when recording is live. Make it clear for which patient the clinician is recording and keep the clinician within the context of the same patient's chart.
Enable multiple ambient recordings. Implement a recording button that enables the clinician to start and stop ambient recording multiple times for the same patient and manage their patient schedule in any order. For example, the clinician should be able to start recording on one patient, stop that recording, work on another patient, and then come back at any time. Dragon Copilot returns content from all ambient recordings combined. We don't recommend implementing a pause button or requiring a 'done' button.
Indicate the cumulative length of all recordings combined. There's a maximum recording length, which continues to increase as the technology improves. Display a list of recordings that have been captured for a patient, including the date and timestamp. This confirms to the user that recordings were successful and shows how much time was spent.
Handle audio interruptions. When the recording is interrupted for any reason (phone call, video app, etc.) inform the clinician that the recording was interrupted and resume recording automatically as soon as possible. If it's not possible to resume automatically, inform the clinician and provide a way to resume recording.
Provide responsible AI disclaimers. It's important to reinforce responsible AI practices by notifying clinicians that all AI outputs can contain mistakes and require review.
Dragon Copilot core functionality
Here we list the functionality provided by the Dragon Copilot web UI, which is available to partners using Dragon Copilot Embedded for Mobile and Dragon Copilot Embedded for Desktop. We recommend enabling the full suite of features for your users.
| Feature | UX best practice |
|---|---|
| Language selection | Enable users to switch languages. |
| Automated styling | Enable users to automatically apply a set of pre-selected styles and templates to their documents. |
| Note | Clinicians prefer having access to content across all devices. This way, they can reference and review content throughout their workflow. They need to know the status of their content, including timestamps, uploading and generation status, version, if there’s any issue or degradation of quality, and if they’ve edited the narrative output directly. The note can and should be available both in Dragon Copilot—such that all available features can be used—and synced with the EHR. |
| Orders and discrete data | Clinicians prefer having access to orders across all devices. This way, they can see orders outputs throughout their workflow. They need to know the status of their orders content, including timestamps, uploading and generation status, type, version, if there’s any issue or degradation of quality, and if they’ve edited the orders output directly. Enable clinicians to view by chronology or order type. Automatically deliver and sync with the EHR. |
| Feedback | It's a requirement of Responsible AI that feedback can be submitted in app for every AI output, including for notes, orders, and copilot responses. Clinicians can rate and comment on the AI output one or more times. Ensure that the feedback is returned with the specific AI content it references. It must be clear where and how to provide feedback. This is a critical input to improving the AI output so feedback should not be subtle. Preferably, feedback should be sent both from the EHR where content is delivered and in Dragon Copilot. |
| Versioning | Allow clinicians to roll back to a prior version of the note, such as to remove an ambient recording from being processed when creating the note, or to revert edits made. Recorded content will still be available in the transcript. Rolling back a note version prevents further ambient recording. |
| Transcript | Give clinicians the speech to text transcript of exactly what was recorded. They need to know the status of the transcript delivery, including timestamps, uploading and generation status, and if there’s any technical issue. Clinicians reference the transcript for a variety of reasons, including as a memory aid, to copy from directly when editing, and as a source of information when reviewing the AI outputs. |
| Prompt input field | Clinicians can ask questions, edit and append content, and create custom AI outputs. They are provided, and can add to, a library of prompts. They can review and undo AI-suggested changes and provide feedback on them. |
| Voice dictation and keyboard editing | Clinicians can also edit directly using voice dictation or keyboard. |
| Voice commands | Clinicians can use Dragon commands to navigate and control their experience. |
| Content delivery | We recommend that ambient AI outputs (note, orders) are delivered automatically to the EHR upon creation. Content from the first recording should append content that was already present in the EHR. Recordings are aggregated together, resulting in delivery of updated ambient content. Once a clinician edits, new content can either be appended or blocked from delivery (merge capabilities aren't yet available). Some EHRs might choose to require the clinician to indicate they're done recording and editing with DAX Copilot before content is delivered to the EHR, but this is not recommended compared to a solution that syncs and merges automatically. |