Getty Images/iStockphoto
AI Medical Scribe Deepens Epic EHR Integration to Enhance Workflows
Improvements to the EHR integration include tools to customize AI medical scribe notes and compatibility with Epic SmartData elements.
Medical scribe vendor DeepScribe has announced several updates to its Epic EHR integration.
DeepScribe's Customization Studio, which allows end-users to build custom models to support organizational needs, is now fully integrated with Epic's SmartData elements.
The vendor's ambient AI scribe captures patient-provider conversations, converts them into customizable documentation within seconds, and uploads it directly to Epic's EHR platform.
Epic and DeepScribe customer Covenant HealthCare helped shape improvements to DeepScribe's integration through feedback from clinicians, technical experts, and organizational stakeholders. Already, the integration is helping Covenant providers reduce after-hours EHR documentation by up to 75 percent.
"We worked tirelessly with DeepScribe developers to enhance the integration in a way that meets clinical needs more holistically," Frank Fear, CIO of Covenant HealthCare, said in a press release. "The primary objective for us was to help DeepScribe establish compatibility with Epic's SmartData elements."
"With that functionality built out, we can now properly leverage all of DeepScribe's unique features in a really cohesive manner," Fear said. "That seamlessness between platforms is empowering us to streamline workflows and improve the clinician-patient experience."
Key features of the integration include:
- Customization Studio: The suite of tools allows users to customize note wording, formatting, and layout.
- SmartData compatibility: DeepScribe pushes information into Epic's SmartData elements, meaning preferences created through the Customization Studio are present within Epic.
- HEAL LLM: DeepScribe's proprietary large language model (LLM), HEAL, allows DeepScribe to deliver accuracy rates 32 percent better than platforms running GPT-4 as a standalone LLM.
- Continuous training: In addition to constant model refinement, DeepScribe uses feedback and any edits made to the note in the Portal to learn and improve its performance over time.
- Visit type templating: DeepScribe will configure clinical notes based on the visit type specified in a provider's schedule within the EHR.
- Plan by ICD10: DeepScribe will enhance its Plan by Problem feature to include corresponding ICD10 billing codes.