Ear Echo Stream

Inspiration

Concussions are often missed in youth sports. A second hit can make things worse. We saw a gap at the sideline where time and tools are limited. Recent research shows that earbuds and a phone can capture inner ear echoes with useful signal. That insight unlocked a fast screen that a coach can run in the field. Our aim is simple. Find likely cases early and send the player to care.

What it does

Ear Echo Stream runs a fifteen second screen. The app plays two tones, records the earbud mic, and measures echo strength against noise. All math runs in the browser. No raw audio leaves the device by default. The result is a clear traffic light. Green means no concern based on this screen. Yellow means repeat or monitor. Red means stop play and refer. A history view shows trends across the season.

How we built it

We ship a pure web app with React and Material UI. Web Audio API generates tones and reads the mic through getUserMedia. An AudioContext path handles timing. A fast Fourier transform in WebAssembly extracts the spectrum. We compute signal to noise in target bands and compare to the player baseline in IndexedDB. An optional export maps results to a FHIR Observation for clinic review. We began with a quick Streamlit proof, then moved to this browser build for speed, polish, and offline use.

Challenges

Audio precision. Bluetooth mics limit the high band. For early tests we guide users to a wired headset and keep tones within the device band. Signal isolation. A poor seal and loud rooms hide the echo. We added an ambient noise gate and a seal check before any capture. Medical validation. A screen must be clear and careful. We show plain language, record consent, and label results as a screen only. Hardware variability. Laptops and earbuds differ a lot. We detect sample rate and codec and log device data to tune the path.

Accomplishments

We turned a clinic style signal into a browser app that runs on school hardware. The kit cost drops from thousands to tens. The UI looks like real medical software and stays simple for coaches. Processing is local by default which protects privacy and works offline. A clean data model is ready for research and future trials.

What we learned

Consumer mics can deliver stable spectra when the seal is good and the room is quiet. Signal processing choices matter more than model size at this stage. Clear copy and large touch targets reduce user error more than any extra feature. Baselines help a lot in the field. Health data needs standard formats from day one.

What is next

Better seal checks with visual prompts and auto retry. Multi band tones and device aware timing checks. Player accounts with season baselines and role based access for staff and parents. Partner schools for a pilot and a formal comparison against common sideline screens. A study plan that records FHIR data and prepares for clinic review. Later, a mobile build if schools ask for it, plus model training on real world data with strict consent. Global use is possible since the stack is light and the app runs offline.

Purpose and Intended Users

Ear Echo Stream is designed for youth athletes, school coaches, and onsite health staff. It addresses the critical gap in early concussion detection by providing an instant screening tool that anyone can use. For athletes, this means a chance to catch a brain injury before it worsens. For coaches, it means no longer relying solely on subjective observation. For school nurses, it provides consistent records that track changes over time. The tool brings clinical insight to the sideline without needing a specialist.

Impact and Success Metrics

There are an estimated 3.8 million sports-related concussions in the U.S. each year, and studies show that nearly 50% of youth concussions go undiagnosed on the day of the injury. Our goal is to make screening accessible to at least 500,000 athletes in the first rollout year. To measure success, we will track the number of completed tests, the percent of red flag results that are referred for medical evaluation, and the average time between injury and screening.

Fit to Theme

The project fits directly within the theme of “Health Tech for Good.” It delivers high-impact brain health screening using only a web browser and standard earbuds with inbuilt microphone. Because all processing happens locally, even schools without strong internet connections can still use the tool. This makes it an equalizer for underserved communities and a true example of accessible, preventative health tech.

User Interface and Flow

Users begin on a home screen with a clean microphone selection and a “Start Test” button. Once the test begins, they go through a seal check to ensure proper earbud placement. A live visualization then displays while the test runs for 15 seconds. After processing, users receive a red, yellow, or green result based on ear echo response quality. Historical test data is accessible from the dashboard, making it easy for nurses or guardians to follow up.

Architecture

The system is built with a strealit frontend. Audio generation and capture is handled with the Web Audio API. For real-time processing, an AudioWorklet node streams data into a WebAssembly-powered Fast Fourier Transform (FFT) module. Baseline data and test results are stored locally via IndexedDB, and optional export functionality integrates with electronic health records through SMART on FHIR protocols. All computation runs in-browser, ensuring low latency and privacy.

Features and Functionality

The current version supports a complete 15-second test flow with signal calibration, echo detection, and noise filtering. Users can compare against baseline metrics, view past results in a trend graph, and choose whether to export data. A hardware guide assists users in choosing the right headset, and consent features ensure compliance for youth users.

Limits and Risks

Like any audio-based tool, Ear Echo Stream can be affected by loud environments. We mitigate this with a noise gating algorithm and environment prompts. Improper earbud placement can affect test accuracy, which is addressed through a seal check routine. Bluetooth microphones may lack bandwidth for full fidelity; we provide setup guidance to recommend wired options.