/metriport

Metriport is an open-source universal API for healthcare data.

Primary LanguageJavaScriptOtherNOASSERTION

Logo

Metriport helps digital health companies access and manage health and medical data, through a single open source API.
Learn more »

Docs · NPM · Developer Dashboard · Website

API Status Check Github Stars License Commits-per-month LinkedIn YC

Support us on Product Hunt and Launch YC

Metriport - Health Devices API - Open-source Plaid for healthcare data | Product Hunt Launch YC: Metriport - Universal API for Healthcare Data

Overview

wearables

Security and Privacy

Metriport is SOC 2 and HIPAA compliant. Click here to learn more about our security practices.

Health Devices API

Our Health Devices API, allows you to gain access to your users’ health data from various wearables, RPM devices, and mHealth sources through a single standardized API.

Out of the box, our Health Devices API supports the following integrations:

  • Dexcom
  • Fitbit
  • Garmin
  • Oura
  • Whoop
  • Withings
  • Cronometer
  • Apple Health
  • Google Fit

...with many more integrations on the way! If there’s an integration you need that’s not currently on here, feel free to shoot us an email and let us know so we can build it, or feel free to fork our code and add the integration yourself.

wearables

Medical API (Coming Soon)

Open-source with native FHIR support. More info on our Medical API here: https://metriport.com/medical/

Getting Started

Check out the links below to get started with Metriport in minutes!

Repo Rundown

API Server

Backend for the Metriport API.

Connect Widget

Pre-built app that you can embed your own app! Use it to allow your users to authenticate with various data sources, allowing you to pull their health data from those sources.

connect widget

Infrastructure as Code

We use AWS CDK as IaC.

Docs

Our beautiful developer documentation, powered by mintlify ❤️.

Packages

Checkout our packages in /pkgs to help you turbocharge your development:

npm

Our npm packages are available in /packages:

iOS

Our iOS packages are available in our iOS repo:

  • Metriport iOS: SDK to integrate with the Metriport Connect Widget and Apple Health on iOS.

Code Examples

Some example projects that serve as examples for how to integrate with Metriport on various platforms - such as iOS and Android.


Prerequisites

Before getting started with the deployment or any development, ensure you have done the following:

  1. Install the prerequisite programs:
  2. Create an AWS account.
  3. Create an AWS IAM admin user.
  4. Setup AWS Route 53 to handle the DNS for your domain, and create a hosted zone.
  5. Follow modules 1 & 2 of this guide for Typescript to bootstrap the AWS CDK on your local machine.
  6. 🥳 🎉 🥳 🎉 🥳 🎉

Local Development

Monorepo

This monorepo uses npm workspaces to manage the packages and execute commands globally.

But not all folders under /packages are part of the workspace. To see the ones that are, check the root folder's package.json under the workspaces section.

To setup this repository for local development, issue this command on the root folder:

$ npm run install-deps # only needs to be run once
$ npm run build # packages depend on each other, so its best to build/compile them all

Useful commands:

  • npm run test: it executes the test script on all workspaces;
  • npm run typecheck: it will run typecheck on all workspaces, which checks for typescript compilation/syntax issues;
  • npm run lint-fix: it will run lint-fix on all workspaces, which checks for linting issues and automatically fixes the issues it can;
  • npm run prettier-fix: it will run prettier-fix on all workspaces, which checks for formatting issues and automatically fixes the issues it can;

Semantic version

This repo uses Semantic Version, and we automate the versioning by using Conventional Commits.

This means all commit messages must be created following a certain standard:

<type>[optional scope]: <description>
[optional body]
[optional footer(s)]

To enforce commits follow this pattern, we have a Git hook (using Husky) that verifies commit messages according to the Conventional Commits - it uses commitlint under the hood (config).

Accepted types:

  • build
  • chore
  • ci
  • docs
  • feat
  • fix
  • perf
  • refactor
  • revert
  • style
  • test

Scope is optional, and we can use one of these, or empty (no scope):

  • api
  • widget
  • infra

The footer should have the ticket number supporting the commit:

...
Ref: #<ticket-number>

Commitizen

One can enter the commit message manually and have commitlint check its content, or use Commitizen's CLI to guide through building the commit message:

$ npm run commit

In case something goes wrong after you prepare the commit message and you want to retry it after fixing the issue, you can issue this command:

$ npm run commit -- --retry

Commitizen will retry the last commit message you prepared previously. More about this here.

API Server

First, create a local environment file to define your developer keys, and local dev URLs:

$ touch packages/api/.env
$ echo "LOCAL_ACCOUNT_CXID=<YOUR-TESTING-ACCOUNT-ID>" >> packages/api/.env
$ echo "API_URL=http://localhost:8080" >> packages/api/.env
$ echo "CONNECT_WIDGET_URL=http://localhost:3001/" >> packages/api/.env
$ echo "CRONOMETER_CLIENT_ID=<YOUR-ID>" >> packages/api/.env
$ echo "CRONOMETER_CLIENT_SECRET=<YOUR-SECRET>" >> packages/api/.env
$ echo "DEXCOM_CLIENT_ID=<YOUR-KEY>" >> packages/api/.env
$ echo "DEXCOM_CLIENT_SECRET=<YOUR-SECRET>" >> packages/api/.env
$ echo "FITBIT_CLIENT_ID=<YOUR-KEY>" >> packages/api/.env
$ echo "FITBIT_CLIENT_SECRET=<YOUR-SECRET>" >> packages/api/.env
$ echo "GARMIN_CONSUMER_KEY=<YOUR-KEY>" >> packages/api/.env
$ echo "GARMIN_CONSUMER_SECRET=<YOUR-SECRET>" >> packages/api/.env
$ echo "GOOGLE_CLIENT_ID=<YOUR-KEY>" >> packages/api/.env
$ echo "GOOGLE_CLIENT_SECRET=<YOUR-SECRET>" >> packages/api/.env
$ echo "OURA_CLIENT_ID=<YOUR-KEY>" >> packages/api/.env
$ echo "OURA_CLIENT_SECRET=<YOUR-SECRET>" >> packages/api/.env
$ echo "WHOOP_CLIENT_ID=<YOUR-KEY>" >> packages/api/.env
$ echo "WHOOP_CLIENT_SECRET=<YOUR-KEY>" >> packages/api/.env
$ echo "WITHINGS_CLIENT_ID=<YOUR-SECRET>" >> packages/api/.env
$ echo "WITHINGS_CLIENT_SECRET=<YOUR-SECRET>" >> packages/api/.env
$ echo "FHIR_SERVER_URL=<FHIR-SERVER-URL>" >> packages/api/.env # optional

Additionally, define your System Root OID. This will be the base identifier to represent your system in any medical data you create - such as organizations, facilities, patients, and etc.

Your OID must be registered and assigned by HL7. You can do this here.

By default, OIDs in Metriport are managed according to the recommended standards outlined by HL7.

$ echo "SYSTEM_ROOT_OID=<YOUR-OID>" >> packages/api/.env

These envs are specific to Commonwell and are necessary in sending requests to their platform.

$ echo "CW_TECHNICAL_CONTACT_NAME=<YOUR-SECRET>" >> packages/api/.env
$ echo "CW_TECHNICAL_CONTACT_TITLE=<YOUR-SECRET>" >> packages/api/.env
$ echo "CW_TECHNICAL_CONTACT_EMAIL=<YOUR-SECRET>" >> packages/api/.env
$ echo "CW_TECHNICAL_CONTACT_PHONE=<YOUR-SECRET>" >> packages/api/.env
$ echo "CW_GATEWAY_ENDPOINT=<YOUR-SECRET>" >> packages/api/.env
$ echo "CW_GATEWAY_AUTHORIZATION_SERVER_ENDPOINT=<YOUR-SECRET>" >> packages/api/.env
$ echo "CW_GATEWAY_AUTHORIZATION_CLIENT_ID=<YOUR-SECRET>" >> packages/api/.env
$ echo "CW_GATEWAY_AUTHORIZATION_CLIENT_SECRET=<YOUR-SECRET>" >> packages/api/.env
$ echo "CW_MEMBER_NAME=<YOUR-SECRET>" >> packages/api/.env
$ echo "CW_MEMBER_OID=<YOUR-SECRET>" >> packages/api/.env
$ echo "CW_ORG_MANAGEMENT_PRIVATE_KEY=<YOUR-SECRET>" >> packages/api/.env
$ echo "CW_ORG_MANAGEMENT_CERTIFICATE=<YOUR-SECRET>" >> packages/api/.env
$ echo "CW_MEMBER_PRIVATE_KEY=<YOUR-SECRET>" >> packages/api/.env
$ echo "CW_MEMBER_CERTIFICATE=<YOUR-SECRET>" >> packages/api/.env

Optional analytics reporting

The API server reports analytics to PostHog. This is optional.

If you want to set it up, add this to the .env file:

$ echo "POST_HOG_API_KEY=<YOUR-API-KEY>" >> packages/api/.env

Optional usage report

The API server reports endpoint usage to an external service. This is optional.

A reachable service that accepts a POST request to the informed URL with the payload below is required:

{
  "cxId": "<the account ID>",
  "cxUserId": "<the ID of the user who's data is being requested>"
}

If you want to set it up, add this to the .env file:

$ echo "USAGE_URL=<YOUR-URL>" > packages/api/.env

Finalizing setting up the API Server

Then to run the full back-end stack, use docker-compose to lauch a Postgres container, local instance of DynamoDB, and the Node server itself:

$ cd packages/api
$ npm run start-docker-compose

...or, from the root folder...

$ npm run start-docker-compose -w api

Now, the backend services will be available at:

  • API Server: 0.0.0/0:8080
  • Postgres: localhost:5432
  • DynamoDB: localhost:8000

Another option is to have the dependency services running with docker compose and the back-end API running as regular NodeJS process (faster to run and restart); this has the benefit of Docker Desktop managing the services and you likely only need to start the dependencies once.

$ cd packages/api
$ npm run start-dependencies # might be able run it once
$ npm run dev

Database Migrations

The API Server uses Sequelize as an ORM, and its migration component to update the DB with changes as the application evolves. It also uses Umzug for programatic migration execution and typing.

When the application runs it automatically executes all migrations located under src/sequelize/migrations (in ascending order) before the code is atually executed.

If you need to undo/revert a migration manually, you can use the CLI, which is a wrapper to Umzug's CLI (still under heavy development at the time of this writing).

It requires DB credentials on the environment variable DB_CREDS (values from docker-compose.dev.yml, update as needed):

$ export DB_CREDS='{"username":"admin","password":"admin","dbname":"db","engine":"postgres","host":"localhost","port":5432}'

Run the CLI with:

$ npm i -g ts-node # only needs to be run once
$ cd packages/api
$ ts-node src/sequelize/cli

Alternatively, you can use a shortcut for migrations on local environment:

$ npm run db-local -- <cmd>

Note: the double dash -- is required so parameters after it go to sequelize cli; without it, parameters go to npm

Umzug's CLI is still in development at the time of this writing, so that's how one uses it:

  • it will print the commands being sent to the DB
  • followed by the result of the command
  • it won't exit by default, you need to ctrl+c
  • the command up executes all outstanding migrations
  • the command down reverts one migration at a time

To create new migrations:

  1. Duplicate a migration file on ./packages/api/src/sequelize/migrations
  2. Rename the new file so the timestamp is close to the current time - it must be unique, migrations are executed in sorting order
  3. Edit the migration file to perform the changes you want
    • up add changes to the DB (takes it to the new version)
    • down rolls back changes from the DB (goes back to the previous version)

Additional stuff

To do basic UI admin operations on the DynamoDB instance, you can do the following:

$ npm install -g dynamodb-admin # only needs to be run once
$ npm run ddb-admin # admin console will be available at http://localhost:8001/

To kill and clean-up the back-end, hit CTRL + C a few times, and run the following from the packages/api directory:

$ docker-compose -f docker-compose.dev.yml down

To debug the backend, you can attach a debugger to the running Docker container by launching the Docker: Attach to Node configuration in VS Code. Note that this will support hot reloads 🔥🔥!

Connect Widget

To run the Connect Widget:

$ cd packages/connect-widget
$ npm run start # available on port 3001 by default

...or, from the root folder...

$ npm run start -w connect-widget

To debug the Connect Widget, you can run a Chrome window by launching the Run Chrome configuration in VS Code.

Utils

The ./packages/utils folder contains utilities that help with the development of this and other opensource Metriport projects:

Check the scripts on the folder's package.json to see how to run these.


Tests

Unit tests can be executed with:

$ npm run test

To run integration tests, make sure to check each package/folder README for requirements, but in general they can be executed with:

$ npm run test:e2e

Self-Hosted Deployments

API Key Setup

Most endpoints require an API Gateway API Key. You can do it manually on AWS console or programaticaly through AWS CLI or SDK.

To do it manually:

  1. Login to the AWS console;
  2. Go to API Gateway;
  3. Create a Usage Plan if you don't already have one;
  4. Create an API Key;
    • the value field must follow this pattern: base 64 of "<KEY>:<UUID>", where:
    • KEY is a random key (e.g., generated with nanoid); and
    • UUID is the customer ID (more about this on Initialization)
  5. Add the newly created API Key to a Usage Plan.

Now you can make requests to endpoints that require the an API Key by setting the x-api-key header.

Environment Setup

  1. Install AWS CLI and authenticate with it.

  2. You'll need to create and configure a deployment config file: /infra/config/production.ts. You can see example.ts in the same directory for a sample of what the end result should look like. Optionally, you can setup config files for staging and sandbox deployments, based on your environment needs. Then, proceed with the deployment steps below.

  3. Configure the Connect Widget environment variables to the subdomain and domain you'll be hosting the API at in the config file: packages/connect-widget/.env.production.

Deployment Steps

  1. First, deploy the secrets stack. This will setup the secret keys required to run the server using AWS Secrets Manager and create other infra pre-requisites. To deploy it, run the following commands (with <config.stackName> replaced with what you've set in your config file):
$ ./packages/scripts/deploy-infra.sh -e "production" -s "<config.secretsStackName>"
  1. After the previous steps are done, define all of the required keys in the AWS console by navigating to the Secrets Manager.

  2. Then, to provision the infrastructure needed by the API/back-end execute the following command:

$ ./packages/scripts/deploy-infra.sh -e "production" -s "<config.stackName>"

This will create the infrastructure to run the API, including the ECR repository where the API will be deployed at. Take note of that to populate the environment variable ECR_REPO_URI.

  1. To deploy the API on ECR and restart the ECS service to make use of it:
$ AWS_REGION=xxx ECR_REPO_URI=xxx ECS_CLUSTER=xxx ECS_SERVICE=xxx ./packages/scripts/deploy-api.sh"

where:

  • ECR_REPO_URI: The URI of the ECR repository to push the Docker image to (created on the previous step)
  • AWS_REGION: The AWS region where the API should be deployed at
  • ECS_CLUSTER: The ARN of the ECS cluster containing the service to be restarted upon deployment
  • ECS_SERVICE: The ARN of the ECS service to be restarted upon deployment

After deployment, the API will be available at the configured subdomain + domain.

  1. Finally, to self-host the Connect widget, run the following:
$ ./packages/scripts/deploy-infra.sh -e "production" -s "<config.connectWidget.stackName>"

This will create the infrastructure to run the Connect Widget, including the S3 Bucket to host the files and the CloudFront distribution. Take note of that to populate the environment variables S3_BUCKET and CF_DISTRIB_ID.

  1. And the following, to deploy the Connect Widget's files on S3 and invalidate the cache on CloudFront:
$ S3_BUCKET=xxx CF_DISTRIB_ID=xxx ./packages/scripts/deploy-widget.sh

Note: if you need help with the deploy-infra.sh script at any time, you can run:

$ ./packages/scripts/deploy-infra.sh -h

License

Distributed under the AGPLv3 License. See LICENSE for more information.

Copyright © Metriport 2022-present