Introduction
The Nabla Core API provides advanced tools to transcribe medical encounters, generate structured clinical notes, extract FHIR-normalized data, and more β all optimized for clinical workflows and seamless integration.
Whether youβre building a telehealth platform, an EHR, or a voice-enabled documentation tool, the Nabla Core API is designed to help you go from raw audio to structured data with minimal effort.
Core capabilitiesβ
π©Ί Medical encounter transcriptionβ
Transcribe doctorβpatient conversations using a medical-grade speech-to-text engine, fine-tuned to recognize terminology like medications, symptoms, and diagnoses.
- For real-time use cases: Use the WebSocket Transcribe endpoint.
- For recorded files (β€10 minutes): Use the REST Transcription endpoint.
- For longer files (β€60 minutes): Use the asynchronous REST endpoint.
π Clinical note generationβ
Transform a transcript into a well-structured clinical note using medical templates like SOAP, APSO, or WCC.
- Supports synchronous and asynchronous workflows.
- Customize per-section content, format (paragraphs vs bullet points), and level of detail.
- Add patient context or enrich an existing note incrementally.
βοΈ Edit note with instructionsβ
Also known as Magic Edit, this feature lets you refine a Nabla-generated clinical note using natural language instructions.
- Provide short, human-readable guidance like "Make note more concise" or "Add more details in the Plan section".
- The original transcript and note are used as context to ensure accurate and relevant edits.
- Great for last-mile customization or for incorporating provider feedback post-generation.
π Patient-friendly instructionsβ
Generate post-visit summaries written in patient-friendly language.
- Supports pediatric contexts via
recipient_type
(e.g., "PARENT"). - Multi-language support with localized output.
π§ Normalize structured dataβ
Extract FHIR-compatible structured data from a Nabla-generated clinical note:
- Extract conditions with ICD-10 codes.
- Capture family history and patient observations with LOINC codes.
- Output aligns with EHR-ready formats for seamless integration.
ποΈ Medical dictationβ
Accurately transcribe dictated speech in real-time, optionally including dictated punctuation:
- Supports both real-time (WebSocket) and asynchronous use cases.
- Ideal for physician notes, documentation, and internal summaries.
β‘οΈ Dot Phrasesβ
Automatically detect and suggest Dot Phrases β provider-defined macros that expand into structured sections during note generation.
- Triggered based on transcript content (e.g., if the provider says "Your physical exam is normal").
- Each Dot Phrase defines a trigger and a replacement block of text.
- Optionally, Dot Phrases can auto-adapt based on the encounter context:
- If
content_auto_update_enabled
is set totrue
, the suggested content will be dynamically refined based on the transcript. - For example, the engine may adjust a standard exam note to mention exceptions or emphasize relevant findings from the encounter.
- If
π Custom dictionaryβ
Add user-specific terms to improve transcription accuracy in specialized contexts.
- Useful for names of organizations, local procedures, or uncommon acronyms specific to your workflows.
- Helps ensure consistent recognition of expressions that arenβt part of the default medical vocabulary.
π³οΈ Report feedbackβ
Submit structured feedback on generated content to the support team for review.
- Report issues with clinical notes, patient instructions, or normalized data.
- Optionally include corrected output and comments from the provider.
- This helps us automate user feedback and ensure the API maintains the highest possible quality.
- If a
reply_to
email is provided, the support team may reach out for clarification or follow-up.
Sample app
To see an example of the API in action, check out our sample-app repository for real-world examples (JavaScript).