About

Why we built it.

Two patients with the same grade 2 murmur. One was audible at the door; the other needed a quiet room and firm pressure. The difference was the chest wall.

17.2 mm 38.9 mm collar body PA12, 45 g collar M40 x 0.75 gasket 60 Shore A
Why a physical part

Software cannot know the wall thickness.

The wall sits between the heart and every microphone on the skin. Its effect appears in the recording, so the insert measures it there and divides it out.

Why it is not a model

One division, no training set.

A learned correction would need paired recordings and imaging from thousands of patients, and its errors would be hard to attribute. This is arithmetic, and it can be checked with a calculator.

Where it is now

Rev D

Printed, hand fitted, workable. More than 100 people have handled it.

Bench

Simulation and bench evidence, 111 recordings. The study has not started.

Tooling

Not paid for, and not dated. What exists is a printed part and a method.

The people who built it.

Engineering, clinical protocol and review. Everyone here has handled the device or read the analysis.

AW

Amy Wang

Founder, system design

Acoustic path, correction, app and the manufacturing envelope.

JW

Jiahe Wang

Co-founder, clinical protocol

MLPAO-registered medical laboratory technician. Owns how a recording is taken.

MK

Manolis Kellis

Advisor, MIT

Signal and statistical methodology: what can be claimed from a recording.

HH

Dr. He Huan

Clinical advisor

Reviews the findings language and the recording protocol.

YD

Dr. Yu Dongping

Clinical advisor

Reviews the site-of-recording taxonomy and what the app should refuse to report.

GY

Grace Yu

Advisor, interface

Reviews the interface and the language of the findings.