Feelix by Sonavi Labs

by Brandon Dottin-Haley

Coauthors: If considered for presentation, Sonavi Labs would prefer to present on 9/22/21 as that is best for the teams schedule. Presenting on behalf of Sonavi Labs will be Ilene Busch-Vishniac, PhD

Medical Devices & Digital Health


Sonavi Labs is on a mission to improve patient outcomes and decrease respiratory disease morbidity and mortality globally by harnessing the most advanced acoustic technology available coupled with clinically proven diagnostic AI software.

There are no tools available that can objectively analyze respiratory abnormalities leaving patients with chronic diseases to use subjective measures to evaluate health trends. Over a quarter of healthcare costs are related to emergency visits and providers lack the tools necessary to effectively track patient trends and deliver appropriate interventions. Chronic respiratory patients need a way to understand their health trends and better communicate with care teams to avoid exacerbations.

After years of research at Johns Hopkins University, Sonavi Labs was formed to deploy Feelix, a platform that features proprietary hardware embedded with clinically validated diagnostic software capable of detecting respiratory diseases like pneumonia and TB in seconds.

The Feelix system also features supportive apps and an integrated cloud platform to provide remote monitoring solutions for chronic disease patients with asthma and Cystic Fibrosis. Feelix is the world’s first diagnostic stethoscope that can be used by clinicians of all skill levels and patients or their caregivers anywhere.

Feelix devices harness the most advanced acoustic technology available coupled with proprietary classification software embedded directly on to the devices. An adaptive noise suppression algorithm uses data collected from 6 on-board microphones to identify body sounds and filter out ambient noise, adapting to changing noise in the environment. The Feelix device is based on AI and Machine Learning as well as a straight computational approach to dynamic (acoustic) noise suppression. The small, handheld device incorporates an FPGA and microprocessor to instantly process data at the edge, allowing Feelix devices to function with or without internet connectivity. The devices are ideal for improving upon current telehealth visits that are currently limited to video conferencing and would allow clinicians to hear patient body sounds in real-time.

Recordings can be stored on the device and transmitted via Bluetooth in real-time or asynchronously. Data from the recordings is stored in our secured cloud and helps clinicians track high-risk patients and deliver interventional support before patients experience an exacerbation. The Feelix system objectively quantifies respiratory abnormalities, enabling providers to recognize when patients are at risk of an exacerbation. Providing care teams with the ability to identify high-risk patients and provide educational guidance can save the US health system billions of dollars each year.

The embedded classification algorithm analyzes the acoustic signals in real-time to differentiate normal sounds from abnormal sounds, then classify the abnormalities to provide a diagnosis. The acoustic quality delivered by the Feelix devices allows the classification algorithms to have high accuracy because computations are not corrupted by poor signal quality.

Sonavi Labs aims to build research partnerships to continue training the Feelix classification algorithm to validate its effectiveness in diagnosing and managing acute and chronic conditions. The team has already achieved a 91% accuracy at identifying pneumonia in pediatric patients and is now developing digital asthma severity monitoring solutions.

Keywords:
Remote Patient Monitoring
Digital Health
Artificial Intelligence
Machine Learning
Acoustics