Feelix by Sonavi Labs

Respiratory diseases affect over 600 million people worldwide, and pneumonia is the leading killer of children under the age of 5. In developing nations such as Bangladesh, respiratory diseases are an epidemic due to the fact that air pollution is increasing and populations are growing, but access to proper diagnosis and subsequent treatment and monitoring are limited. 

Sonavi Labs created the Feelix product line to enable clinicians, global health workers, and patients to quickly identify acute and chronic respiratory problems. Feelix captures lung sounds to analyze and classify through on-board artificial intelligence. Feelix can also connect to mobile phones to provide physicians and patients with trend data and recommendations. For patients in Bangladesh this work will save lives, reduce lost workdays, and improve operational efficiency in clinics, hospitals, and in the field. 

Working in partnership with BUET, we plan to expand into cardiovascular diseases, which cause 31% of global deaths.

The global shortage of medical professionals creates a burden on patients who desperately need timely care, but also on the overworked, under-resourced professionals.  In Bangladesh, there is approximately 1 physician to every 1,800 patients, which is similar to the situation in many of the markets we serve, such as Malawi and Peru. The shortage of physicians exacerbates the impact of treatable conditions like pneumonia, which is responsible for 28% of deaths in children under 5 and is the leading cause of pediatric mortality. Globally, hundreds of millions of people are impacted by respiratory and cardiac diseases, and the burden is only going to increase as the number of cases continues to exceed the available diagnostic resources.

Community health workers (CHW) are often deployed, in countries like Bangladesh.  For many diseases CHWs see, WHO guidelines encourage the over-prescription of antibiotics, so having real-time diagnostic support will reduce the likelihood of creating antibiotic resistance.

We designed the Feelix platform to be user-friendly, informative and rapidly responsive (under 15 seconds) to minimize the need for training while maximizing the impact. Feelix equips CHWs and clinicians with easy-to-use devices that enable them to communicate with each other more effectively and to have greater confidence in their diagnoses.

A profusion of a simple screening tool in areas that do not have direct access to specialists will be a paradigm shift in reducing mortality rates for the respiratory and cardiovascular diseases. Even the slightest change in screening capability could have a significant impact on patients and clinical workers by increasing early detection and enabling long-term monitoring.

We have worked with community health workers (CHWs) in Bangladesh for the past five years through iccdr,b and the Projahnmo study group, and CHWs in Peru and Malawi to collect and analyze pediatric patient data, develop advanced tools, and validate the usability of the device. Our aim was to develop closely with the people who engage with the patients daily to help them conserve time and resources, while improving outcomes. 

We learned from CHWs that it is necessary to have advanced analytics on-board to work regardless of internet connectivity. We also recognized that noise suppression would be needed to operate in non-traditional clinical settings like rural clinics and patients’ homes. 

By integrating Feelix into clinical workflows, we enable healthcare workers of Bangladesh to be more effective and accurate. We believe that placing Feelix in every community will lead to healthier, longer lives.

The Feelix smart stethoscope comes in two versions. The low-cost version features a simple ‘red light, green light’ indication that allows for one-click screening without the need for extensive training to understand body sounds. The premium version, FeelixPro, adds a graphical screen showing real-time analysis results and further clinical information. Trained physicians can listen in to clean auscultation signals and confirm diagnosis against the real-time analysis results while CHWs can screen patients without needing to listen. Our devices have a number of other advantages compared to competing solutions. Feelix is a noninvasive device that is orders of magnitude less expensive than ultrasound and radiographic equipment. Feelix incorporates adaptive noise suppression, making it appropriate for use in nearly any environment. It uses an array of microphones below the diaphragm, making it sensitive over its entire face and less reliant on placement by untrained personnel. The classification algorithm uses a convolutional neural network to analyze body sounds in real time and can correctly identify abnormal cases with 91% accuracy. The device is stand-alone so a mobile phone or computer is not required and will operate without electricity or internet connectivity. All these features fit into a small package: 5cm in diameter and 3cm height.

Bluetooth integrates the devices into a larger platform that comprises mobile apps and a cloud system. The app can store results and integrate with the cloud. We can use artificial intelligence algorithms to understand patient and population trends, make predictions for future events, and provide treatment recommendations. The cloud can also share results with medical professionals instantly or follow-up with patients. This system supports healthcare professionals and expedites more accurate and effective care through the objective analysis. 

A third offering, the Feelix@Home platform, enables patients to manage their chronic conditions like asthma and COPD from the comfort of their home with the accuracy of a professional. Early studies have shown that home monitoring programs increase treatment adherence and decrease emergency visits. Patients are encouraged to take daily recordings of their lung sounds and the platform assesses severity, analyzes trends, and provides educational content to help avoid acute events. Care teams can easily evaluate a patients’ status remotely and engage directly if needed. 

The cloud-based analytics also offer health systems the ability to assess best practices and more strategically distribute resources. The fully integrated platform allows everyone across the spectrum of healthcare to engage in substantive and impactful ways.

Feelix is far more advanced than any other medical-grade, digital respiratory screening device in the world with embedded multiple sensors and advanced algorithms loaded directly onto the hardware to operate in real-time and in a wide variety of environments. The research proposed with BUET will enhance the on-board algorithms to analyze both heart and lung sounds in real-time, in order to assess the patient for any pulmonary or cardio-vascular condition.

Feelix features adaptive noise suppression that removes external sounds in any environment from patient screenings. The devices also include on-board storage that is coupled with Bluetooth connectivity to allow users to record data and share it with a consulting physician or securely upload into an EHR or research database. 

The on-board auto-classification algorithm that has been deployed onto the device works in real-time to identify abnormalities via auscultation, in the same way a physician would. The algorithm has been validated at 91% accuracy, with a 90% sensitivity and 92% specificity.  Further, we can identify wheezes and crackles independently and even grade the types of these sounds.  With this work under our belt, we believe extension to cardiac sounds will follow suit.

Our solution is intended to impact and modernize healthcare from the patient all the way through the healthcare administration system. With enhanced accuracy and immediate detection capabilities in the home of every patient and in the hands of every clinician, treating and managing acute and chronic conditions becomes a holistic process that not only evaluates the patient, but the course of treatment in real-time.

The Feelix platform addresses the needs and engages of all stakeholders within the Bangladeshi healthcare ecosystem, from the patient to the health administrator.

By deploying an objective, dry, digital diagnostic solution across the patient population and into clinical workflows, we can expedite patient treatments with confidence, reduce the need for expensive, environmentally unfriendly, single-use wet diagnostic tests and continue to glean new insights from research based on patient screenings.

Community Health Workers are being asked to perform the work of highly trained physicians in many cases, and, following WHO guidelines, are inadvertently exacerbating the challenge of antimicrobial resistance. We can support CHWs by arming them with smart devices that not only offer treatment guidance in the field, but connect them with centralized health centers. In many ways, this problem is similar to that of parents and caregivers in developed markets who are tasked with caring for family members with chronic conditions.

There are early studies that indicate digital diagnostics can reduce the unnecessary use of resources by up to 56% and remote monitoring solutions can improve outcomes and curb some of the burdens of chronic disease management.

By working with health systems, in Bangladesh and across the world, we can use data optimize treatment plans and personalize health management programs. We understand the importance of empowering patients and their support system with powerful tools that facilitate seamless engagement and rapid alerts. Feelix was designed to save time, save costs, and most importantly, save lives.

We have deployed 100 devices to Community Health Workers in Bangladesh, Malawi and Peru, and to physicians in the United States as well since the inception of our research 5 years ago. By the end of this year, we will have deployed another 100 commercially-ready devices to research programs around the world, including Bangladesh, Belgium and Peru. Our devices have collected thousands of patient screenings, and the auto-classification algorithm is being deployed to support Community Health Workers every day.

We estimate that for every 100 devices that we deploy to global health workers who engage with approximately 20 to 30 patients per day, our devices screen approximately 500,000 patients every year, collecting over 2 million recordings. If we assume the CHWs, armed with Feelix can detect 10% more of otherwise missed cases of pediatric pneumonia, which was the first population with which we validated the technology, then we can potentially save 275 children's lives per day every year.  These are lives that would have been lost to a disease that is detectable and treatable.

Our goal is to enhance our proprietary machine learning algorithm and delve deeply into respiratory diseases and conditions. Working with the BUET team to enhance the Feelix platform to address cardiac conditions, rapidly accelerates and amplifies the technology's ability to support clinicians and patients in a wide variety of settings, ultimately supporting millions, if not billions of patients and their healthcare providers.

Milestones for the next year are to complete the FDA and CE regulatory approval processes to allow mass distribution of our platform and to complete additional clinical trials. Completing the regulatory process will trigger our ability to begin selling the product globally.  Additional clinical data will allow our algorithms to continue improving through deep learning.  This will also permit researchers to study the impacts of age, geography and other characteristics on respiratory illnesses. 

We are most excited to begin additional research and development with Dr. Taufiq Hasan at BUET, who will be leading the charge using our classification algorithm to identify cardiovascular conditions.

In years 2-5 we aim to quantify our impact on clinical workflow efficiency and to introduce a home version of Feelix that is designed for management of chronic diseases such as TB and asthma. Our work with hospital systems to demonstrate workflow impacts of Feelix Is just beginning, with firm results anticipated in year 2 and beyond.  We anticipate that the Feelix platform will decrease the need for expensive and time-consuming lab tests, thus enhancing work efficiency for hospitals and clinics.  The design of a version of Feelix for use at home requires the creation of an appropriate app rather than changes to the hardware and this will occur in years 2 and beyond.

Ultimately, we aim to ensure the Feelix platform is as ubiquitous as the thermometer to maximize our ability to save lives, decrease lost work days, reduce treatment costs, and improve medical outcomes.

Currently, not having received CE marking or FDA approval presents a significant financial barrier for our company because we are not able to generate revenue from sales. Our research and development is currently supported by seed fundraising and grant applications.

The cost of our device might also be a significant barrier in some low and middle-income countries, even though we are orders of magnitude less expensive than ultrasound or radiographic equipment. We have identified multiple components within our hardware that can be optimized for significant cost reduction. Bringing the cost of our device down is a key priority and will be achieved through additional research funding and developmental support.

We have designed our platform to be extremely user friendly, however, technology is often a barrier of entry for some people. Entering multiple markets is also challenging given volatile cultural, political and economic circumstances. We have recognized numerous regions that would significantly benefit from a profusion of our technology but will need to navigate the various nuances that exist when serving these populations.

We are preparing our regulatory submission, which will be submitted in November 2019, and feel very confident in our pathway and the predicate devices identified in our files. Our regulatory consultant has extensive experience in medical devices, and was funded through our partnership with the Consortium for Technology and Innovation in Pediatrics (CTIP), an FDA funded, innovation support program.

We are also building relationships with numerous hospital systems, government agencies, health administrations and global aid organizations to support our entry into developing and frontier markets. FINDdx has incorporated our digital diagnostics platform into their 2020 - 2025 global aid strategy, and will be supporting our entry into low and middle income countries.  We are also in discussion with Philips for inclusion of our devices in their new platform for CHWs.

In Bangladesh, specifically, we are going to leverage our existing relationships within the health system to ensure a smooth and successful collaboration with BUET. Our devices have been developed with the support of data collection initiatives in Bangladesh, and we hope to continue to work with health workers in the field, and within the Ministry of Health to optimize healthcare delivery for all of the people in the country.

We hope that we can use Bangladesh as a model for expanding our impact in other countries. By supporting the health administrators, payers and providers with more effective data and real-time analytics, we can build collaborative relationships that foster better adoption in the community, and smoother deployment on the ground.

We first began collecting research in Bangladesh to develop this technology along with the Bill and Melinda Gates Foundation, through the Pneumonia Etiology Research for Child Health (PERCH) study. Bangladesh was one of the original sites in the study, and since the inception of our research, we have collected thousands of pediatric lung sound recordings that can be used for future research.

Since the company's founding, we have continued to work with physicians and Community Health Workers (CHWs) to refine and enhance our technology. We have sought the input and feedback from the people on the ground who would be using our technology every day, and made sure to include them in every step of development.

Dr. Taufiq Hasan has been building the mHealth lab at BUET for the past two years, and is using his experience in machine learning and bio-medical engineering to enhance the university's academic programs and research capacity.

Our collaboration with Dr. Hasan will accelerate our efforts and potential impact in Bangladesh and the world at large.

There will be approximately 20 individuals collaborating on this project, mostly in Bangladesh collecting and analyzing data.

We anticipate 2 part-time individuals from Sonavi Labs as Principal Investigators and 3 full-time engineers will be needed to annotate data and develop strategies to update the hardware and software.

Approximately 4 individuals from BUET will be involved. 2-3 full-time researchers will analyze data and Dr. Hasan will serve as part-time Principal Investigator.

We will deploy at least 10 full-time Community Health Workers to implement our solution on the ground, provide additional feedback and data collection support, which will help us to continue enhancing the technology.

Feelix is the brain child of the Whiting School of Engineering , and the Bloomberg School of Public Health, proving that diverse academic teams can spark worldwide change. Our leadership team has come together to harness over 50 years of research, engineering, and business experience to commercialize the technology with a focus on making it accessible to populations most in need.

Ian McLane, our Chief Technology Officer, is one of the principal inventors of the AI at the core of Feelix. His doctoral studies at Johns Hopkins focused on electrical engineering, computer science, artificial intelligence, and signal processing.

Chief Innovation Officer, Dr. Ilene Busch-Vishniac, is an expert in acoustics, and is the former dean of the Whiting School of Engineering. She has received numerous patents for her work and continues to deliver talks around the world.

Ellington West, the CEO of Sonavi Labs, brings over a decade of experience in market research, and was the Sales Director for a national healthcare organization.

Brandon Dottin-Haley is a distribution and business development expert with over 15 years experience in customer acquisition, and marketing strategies.

Dr. Taufiq Hasan is a professor of Bio-Medical Engineering at BUET in Bangladesh, and has spearheaded the development of the universities mHealth Research lab since leaving his position as a Research Scientist at Bosch.

This team is uniquely qualified to develop ground-breaking technology and distribute it around the world.

Sonavi Labs is partnering with the mHealth Lab at BUET to expand the features of the Feelix platform and enhance the embedded classification algorithm.

We are partnered with icddr,b to deploy our technology in the field and collect additional data to validate the platform and its impact in global health settings.

We are partnering with FINDdx (The Foundation for Innovation and New Diagnostics) to study Tuberculosis, one of the most fatal respiratory diseases. FINDdx is seeking to incorporate Sonavi Labs digital diagnostic program across the low and middle income countries the organization serves. FINDdx is a WHO partner, and is providing us with support to scale our innovation and optimize it to make the platform more accessible.

Sonavi Labs is a resident of the Johnson and Johnson, JLABS program. As a resident of the program, Johnson and Johnson is facilitating market strategy development and exposing our technology to health administrators, government agencies and global health markets.

Phase Margin has been an instrumental partner in the design and engineering of the Feelix platform, supportive apps and cloud system.

We are partnering with Harbor Designs and Manufacturing, a Baltimore-based procurement and engineering firm that is supporting our production, packaging, and distribution.

We have a multi-pronged business model to provide significant value to our customers, which we delineate into three segments. Our business consists of generating revenue through direct hardware sales, subscription software services, and providing data analytics.

Initial device sales will be targeted towards global health organizations, researchers and hospital systems. Providing diagnostic decision support helps to improve efficiency and accuracy within the treatment regime and reduces administrative burdens. This will also be a significant value to researchers. For example, the University of Antwerp will be implementing our devices to collect data on adult patients with Cystic Fibrosis.

We will continue building partnerships that subsidize the devices in order to deploy them into low resource settings but are also actively working to reduce the cost of production.

Devices will also be made available to patients managing chronic conditions through insurance programs as well as commercially thorough e-commerce platforms. New CPT codes allow for the reimbursement of telehealth services, allowing patients to benefit from less travel and more consistent engagement with their healthcare provider.

Subscription services will offer smaller, private practices and larger hospitals management tools to enable and sustain quality communication between patients and providers. Longitudinal trend analysis also has the potential to reduce the likelihood of emergency incidents by providing more in-depth insights into treatment efficacy.

Providing analytics services to hospital systems, payer organizations and government organizations will also generate revenue to support further research and development and aid with resource allocation decisions.

Sonavi Labs will generate funding through multiple service models, additional grant funding to support further research and an additional capital raise to scale our business across multiple global markets.

Capital will be raised through device sales, subscription services and providing data analytics as a service to health systems, clinicians, and care providers. Providing ongoing data management services to hospitals and patient engagement dashboards for physicians will yield enough revenue to support overhead and significantly contribute to research and development of new value propositions. We will continue to collaborate with organizations around the world to participate in research initiatives.

Additional grants will be sought to advance research initiatives and deploy our technology across low and middle income countries. We are actively seeking additional funding opportunities and have applied to programs such as the Development Innovation Venture (DIV) program, sponsored by USAID. 

To increase scale and capacity of the company, we anticipate completing a Series A funding round after we have completed the regulatory approval process and launched our product.

We have been working in Bangladesh since the start of this project, and this Tiger Challenge presented itself at a serendipitous time our development. We recently met Dr. Taufiq Hasan, and in our first conversation, realized that we could potentially revolutionize disease detection and management. By creating a device that could instantly analyze cardiovascular 

The Feelix platform, in many ways was built on the data we collected in Bangladesh, so we are proud to have the opportunity to continue our research with Dr. Hasan at BUET and create technology that can save lives.

We hope that by deploying the Feelix platform across Bangladesh, we can support patients in remote communities and densely populated urban centers. We considered the people of Bangladesh and their lives in the future to create a solution that will continue to evolve as the country evolves.

The Tiger Challenge is meant to bring teams together to address critical issues. There are millions of people impacted by respiratory diseases, and many will not be able to access the care they need, and we consider that a critical and dire situation that needs urgent and assertive attention. Our team is determined and confident that we have the ability to develop solutions 21st century challenges and create tools that will impact healthcare forever.

We would greatly appreciate the opportunity to partner with global health organizations, specifically Community Health Worker programs and research initiatives. We have been working with icddr,b in Bangladesh for several years and would like to collaborate with other non-profit organizations and federal aid agencies like Dr.s Without Borders and USAID. PATH and the Global Drug Facility would also be great partners in our efforts to distribute the technology and address Tuberculosis. These organizations can support our data collection initiatives and help to deploy our solutions at scale because of their vast reach and global impact.

We would like to connect with hospital systems to deploy our technology to patients and clinicians. Working with organizations like the VA, Mayo Clinic, Massachusetts General, Providence/St. Joseph's, Kaiser Permanente and Johns Hopkins Medical system would allow us to develop strategic modernization initiatives that incorporate key data to optimize best practices. We believe that we can improve workflow efficiency and patient outcomes by deploying the Feelix platform across these and other major hospital systems.

Ideally, we would like to partner with the World Health Organization (WHO) to improve global standards of diagnosis and disease treatment. We believe that data and technology should be a constant thread in healthcare, and the Feelix platform is fully prepared to tackle global health challenges.