Telehealth medical interpretation platform

Built Cloudbreak Health a WebRTC platform that puts patient, provider, and interpreter in one live session.

Industry

Healthcare

Key result

Real-time

Interpretation sessions

Services

Custom Solutions

Healthcare

The Challenge

Healthcare providers see patients who speak hundreds of different languages, and most hospitals cannot staff interpreters for every one of them. In-person interpreters have to be scheduled and have to travel, and sometimes they are not available at all. That is a problem when a patient arrives in an emergency, and it is equally a problem for routine telehealth visits where the patient and the provider do not share a language.

The consequences are clinical, not just logistical. According to the American Journal of Public Health, language barriers contribute to higher rates of medical errors and longer hospital stays. Before the project, interpreters had to be physically present, each session carried a high cost, and providers waited for an in-person interpreter to arrive — delays that landed in the same encounters where accuracy mattered most.

Cloudbreak Health wanted to solve this. They needed a platform where patients, providers, and interpreters could all connect in real time — audio and video — from anywhere. The system had to be reliable. It had to be fast. And it had to work inside a medical context, where communication accuracy is critical. Transmitting patient health information meant HIPAA requirements were baseline constraints on the build rather than something handled afterwards.

The Solution

We built a WebRTC-based application that connects three parties at once — the patient, the healthcare provider, and a remote interpreter — so all three can see and hear each other in real time, with no phone calls and no delays. Off-the-shelf video tools are not built for the complexity of three-party medical interpretation. WebRTC let us build exactly what was needed: a custom session management system with role-based controls, audio routing, and a connection architecture designed for reliability in healthcare settings.

A session broker handles the WebRTC handshake for all three participants. When a provider requests an interpreter, the system finds an available interpreter for the right language and connects all three parties within seconds, each with a separate audio and video stream. A provider can go from requesting an interpreter to being fully connected in under 30 seconds. Language coverage depends on interpreter availability rather than the technology — the platform connects to interpreter networks and is itself language-agnostic.

Audio routing was the harder part. In a normal video call everyone hears everyone; medical interpretation does not work that way. There are moments when the interpreter needs to speak to just the provider, or just the patient. We built configurable audio routing so each session could be managed like a real interpretation, with natural turn-taking and control rather than a single open channel.

Speed was the design constraint throughout. The system was built to minimize latency both in connection setup and in the audio and video streams, because every second matters in a clinical encounter. The connection flow was designed to be as fast as technically possible, with no unnecessary steps between a provider's request and a live three-way session. WebRTC also runs in the browser, so there is nothing for a patient or provider to download before a session starts.

Results

Real-time

Interpretation sessions

WebRTC

Technology platform

Audio + Video

Connection types

Multilingual

Language support

Parties connected per session

Under 30 seconds

Provider request to live session

Before & after

Before

After

Interpreters had to be physically present

Real-time interpretation via video

Delays waiting for in-person interpreters

Immediate connection with interpreters

High cost per interpretation session

Lower cost per session

Language barriers causing medical errors

Accurate communication across languages

Common questions

Off-the-shelf video tools are not built for three-party medical interpretation. WebRTC is a browser-native protocol that establishes direct peer-to-peer connections, which keeps latency low enough that an interpreter can speak almost simultaneously with the provider. It also handles adaptive bitrate, adjusting video quality to available bandwidth to keep the call stable. Building on it allowed custom session management, role-based controls, and audio routing that a conferencing product would not expose.

Visual cues matter in medical settings. A patient's body language, facial expression, and physical condition carry information that audio alone cannot. Interpreters also need to see both patient and provider to interpret accurately, since medical terminology is complex and visual context helps in choosing the right words. Video interpretation reduces miscommunication, which in a medical context has direct consequences for patient safety.

Yes. WebRTC uses end-to-end encryption by default, and combined with proper server-side controls and access management it can be deployed in HIPAA-compliant environments. Any platform transmitting protected health information also needs signed Business Associate Agreements with third-party infrastructure providers, access controls limiting who can join a session, and audit logging of session activity. Those were treated as baseline constraints on this build.

More case studies

See all 18 case studies →

Get started