Introduction
The Nabla Core API allows you to transcribe medical encounters, generate structured clinical notes, extract normalized data, and more!
Main features
Medical encounter transcription
Use our speech-to-text engine, fine-tuned on medical terms (medications, diseases, symptoms, etc.), to transcribe a medical encounter.
- For real-time transcription: Use the Web Socket Transcribe API.
- For small audio files: Use the REST API.
Clinical note generation
Given a transcript of an encounter — either from our transcription API our from another source — you can generate a summarizing clinical note.
For an encounter of about 10 minutes, generating a clinical note from its transcript takes around 20 to 40 seconds. For longer transcripts, the generation will take few more seconds. You can either make a synchronous HTTP call, or rely on our asynchronous webhook callback which avoids blocking calls and is more resilient to network interruptions.
Patient instructions
Summarize, with a patient-friendly language, the instructions given by the healthcare provider during an encounter.
Normalized data from a note
Extract conditions (including the related ICD-10 codes) and family history from a given clinical note, in a FHIR-compatible format.
Medical dictation
Send live audio of a provider's dictation to our speech-to-text engine, and receive accurately transcribed text in real-time.
Custom dictionary
Add expressions that are frequently mistranscribed to improve transcription quality.
Sample app
To see some code examples of how the Nabla Core APIs can be used, check out our sample-app repository (written in JavaScript).
You can download the OpenAPI file for the server API or the OpenAPI file for the user API and use them in tools like Postman or Swagger
(Note that these files do not include websocket APIs).