Connected respiratory diagnostic tool

LUNGS HEALTH
ANALYZER

Category

Digital Health
Medical
IoT

SERVICES

Industrial Design
Hardware engineering
Embedded Firmware
Mechanical engineering

CURRENT STATUS

Clinical tests

The Challenge

In March 2020, the world was hit by the most severe pandemic of the 21st century. Doctors and healthcare professionals were overwhelmed. Patients with mild symptoms needed a new way to check on their respiratory health without a visit to a hospital.

THE OPPORTUNITY

AJProTech partnered with a Venture-backed startup and a number of U.S. Hospitals to create a new precise, non-invasive and radiation-free lung imaging system using acoustic technology.

health health

TRADITIONAL APPROACH

Understanding the location of abnormal lung sounds can help identify areas of pathology as well as assessing the severity of the pathology.

Traditionally, doctor attaches a stethoscope to specific points of patient’s back, moving from one spot to another. Doctor must be involved in the process at all times

PRODUCT OBJECTIVES

  • Capture respiratory sounds in multiple locations simultaneously.
  • Analyze data with Machine Learning algorithm.
  • Eliminate doctor presence during the test.
health health

OUR APPROACH

We approached this project using an agile and iterative process. By combining design, engineering, and rapid prototyping, we developed a technically feasible solution in a short time.

Clinical studies identified 36 specific locations on patient body where acoustic sensors need to be placed.

Digital stethoscopes were selected as optimal sensors to capture respiratory data. Integrating sensors into a vest allowed to hold sensors in place with required force against body.

Custom Controller collected data from sensors and transferred to a computer Software with Machine Learning and data analysis allows to identify abnormal conditions.

HOW IT WORKS

01.
The T-Sense is a multichannel digital stethoscope with 32 microphone transducers fixed over the thorax.
02.
Each transducer records vibrations as air moves in and out of lungs. Data is sent to host PC software through USB.
03.
Machine Learning algorithms analyzes correlations between channels to provide comprehensive report of respiratory health.

What can we create together?

INDUSTRIAL AND MECHANICAL DESIGN

Our vision for T-sense was that it would be used by nurses and medical assistants at first. Nurse puts the wearable device on a patient and device collects all data.

After 30 design versions, we selected a form factor of a vest with an array of sensors on the patient’s back and size adjustment at front. We explored aesthetic design and technical constraints to achieve user-friendly and feasible version.

Each sensor needed to be pressed against patient’s body with equal force to collect proper data. That was achieved by using individual spring mechanism and selection of suitable materials.

ELECTRONICS DESIGN

Proof of Concept was done with off-the-shelf digital stethoscopes connected to PC through USB. However, 32 sensors priced at $400 each were cost-prohibitive and we needed to have a single connection to computer.

We created elegant and robust stethoscopes with microphones placed inside the plastic enclosure and custom membrane. Respiratory sounds are captured by membrane pressed against human body and recorded by the microphone.

Custom-made central controller PCB collects data from all 32 sensors and send to a PC for further analysis through USB. Next version of device may be battery powered and use WiFi connectivity.

PROTOTYPING

We used 3D printing and CNC machining to rapidly prototype mechanical components. We refined mechanical design to provide required force against patient body and optimize assembly process

Custom electronics PCBA were fabricated, assembled, and tested. Device includes a small PCB in each sensor and two controller boards.

Acoustic devices are very sensitive to noise. Using a combination of sound-absorbing materials and microphone membrane design, we significantly improved signal-to-noise ratio.

ACOUSTIC RESULTS AND DATA ANALYSIS

T-sense captures respiratory sounds from 32 locations distributed on patient body. Software analyses precisely synchronized multi-channel data and creates spatial model of patient's lungs.

Machine Learning algorithms then allows assisted or automated diagnosis of various respiratory conditions.

Development of training of ML algo is being done in collaboration with participating hospitals.

Software may also apply various filters to improve SNR.

PILOT AND MARKET VALIDATION

High Level software is being developed. Participating physicians praised device for high accuracy of data.

We produced sufficient number of alfa samples for clinical tests. Each sample was tuned to provide consistent output data.

Impact

Our partners received over $500,000 of funding from Google Accelerator and Y Combinator for further software development in collaboration with participating physicians.

Upon completion of clinical trials, product will go into commercialization phase.

Download this case study

Get a PDF version of case study to download and share.

Contact us

United States
Los Angeles, CA9410 Owensmouth Ave,Chatsworth, CA 91311
Taiwan
ManufacturingNew Taipei City120 Nan Shi St.
Europe
EngineeringComing soon