Quickstart — Coding

Completing this quickstart gets you a working code prediction response from the Medera Coding API. You can then explore the full API and use cases from there.

Prerequisites

A Medera Console account — sign up at the Medera Console.
A Developer API Key — see Creating Clients to generate one with read:phi and write:phi scopes for the appropriate project.

Install the SDK

Choose the SDK for your stack.

npm install @medera/sdk # or pnpm add @medera/sdk

pip install medera

Initialize the client

Replace the placeholders with values from your Console.

import { MederaClient } from "@medera/sdk"; const client = new MederaClient({ apiKey: process.env.MEDERA_API_KEY!, baseUrl: "https://api.medera.info", }); 

from medera import MederaClient import os client = MederaClient(api_key=os.environ["MEDERA_API_KEY"]) 

The SDKs handle token acquisition, refresh, and retries automatically. There is no need to manage bearer tokens manually.

Send a prediction request

POST a clinical note (or a documentId) to /api/therapy/suggest-codes. The code_systems field controls which classifications are applied.| Parameter | Type | Required | Description | | :--- | :--- | :--- | :--- | | code_systems | string[] | Yes | One or more coding systems. Recommended pairs in Coding Systems. | | context | object | Yes | The clinical input: { "documentId": "doc_abc" } or { "text": "..." }. | | min_confidence | number | No | Filter out predictions below this threshold (default 0.5). |

const response = await client.coding.suggestCodes({ code_systems: ["icd-10-cm", "cpt"], context: { documentId: "doc_abc" }, min_confidence: 0.7, }); 

response = client.coding.suggest_codes(code_systems=["icd-10-cm", "cpt"], context={"documentId": "doc_abc"}, min_confidence=0.7,) 

cURL curl -X POST https://api.medera.info/api/therapy/suggest-codes \ -H "X-API-Key: $MEDERA_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "code_systems": ["icd-10-cm", "cpt"], "context": { "documentId": "doc_abc" }, "min_confidence": 0.7 }' 

Read the response

A successful response returns three top-level fields: - codes — the model’s high-confidence predictions. Each entry includes code, display, confidence, anchor (with documentId, span, and quote), and optional severity_qualifier_suggestions. - candidates — lower-confidence codes worth reviewing. - usageInfo — credits consumed for this request.

{ "codes": [ { "system": "icd-10-cm", "code": "F32.1", "display": "Major depressive disorder, single episode, moderate", "confidence": 0.92, "anchor": { "documentId": "doc_abc", "span": { "start": 412, "end": 489 }, "quote": "Patient reports persistent low mood for 6 weeks..." }, "severity_qualifier_suggestions": ["severe"] }, { "system": "cpt", "code": "90834", "display": "Psychotherapy, 45 minutes", "confidence": 0.95, "anchor": { "documentId": "doc_abc", "span": { "start": 12, "end": 38 }, "quote": "45-minute individual psychotherapy" } } ], "candidates": [ { "system": "icd-10-cm", "code": "Z71.1", "display": "Person with feared complaint in whom no diagnosis is made", "confidence": 0.58 } ], "usageInfo": { "creditsConsumed": 1.0 } } 

Combine coding systems

Most US encounters require more than one coding system. Pass multiple values in code_systems — for example, pairing ICD-10-CM with CPT for an outpatient visit. The response includes codes from all requested systems in the same codes array, each tagged with its system.

const response = await client.coding.suggestCodes({ code_systems: ["icd-10-cm", "cpt", "hcpcs"], context: { documentId: "doc_abc" }, }); 

response = client.coding.suggest_codes(code_systems=["icd-10-cm", "cpt", "hcpcs"], context={"documentId": "doc_abc"},) 

See Coding Systems for recommended pairings by encounter type.

Next steps

Code Prediction

Endpoint, supported features, coding system options.

How it works

Request schema, response fields, evidence spans, filtering.

Use cases

Encounter coding, CDI review, revenue cycle.

​Prerequisites

​Next steps

Code Prediction

How it works

Use cases

Prerequisites

Next steps