Creaking joints come from cartilage rubbing on other soft tissue. A growling stomach signals hunger or that you’ve eaten something that disagrees with you. And though a cough can mean that you are ill, you might also be perfectly healthy and just took too big of a gulp of your drink. Even more interesting is that our body sounds and its variation over time could tell us a lot not only about a particular organ (e.g., your knee cartilage and tendons), but also about general wellness, and respiratory, cardiovascular, digestive, and neurological health. This article digs into the future of acoustic epidemiology and our growing ability to interpret the body’s sounds using artificial intelligence (AI).
AI has already made a big impact in medicine in the visual realm. By detecting abnormalities, classifying and quantifying cancerous cells, and assisting surgeons by providing real-time guidance, visual AI has improved early detection, sped up diagnosis, and increased precision and accuracy across a number of medical specialties.
So what about all those noises the body makes? Acoustic AI has been slower to develop but is gaining momentum at a rapid pace, using similar machine learning and neural networks to crunch huge data sets and perform analyses quickly that would take humans centuries.
At Hyfe, our focus is on respiratory health and, in particular, the number one body signal of respiratory distress and illness: cough. Over the past three years, we have assembled the world’s largest cough data set and learned that there is meaning in the sound, frequency, and patterns of coughing as there is in the division of cells within the body. Until recently, so much of the meaning of cough has been hidden because we lacked a way to monitor cough over time, continuously and unobtrusively.
Previously available short-term cough monitoring techniques and devices yield limited or misleading data and are of meager clinical value. The absence of a tool to measure changes in cough effectively over time has also inhibited the development of effective pharmaceuticals. The last time the FDA approved an antitussive was in 1958, the year the hula hoop was invented. The world’s half a billion chronic coughers deserve better.
We are entering the dawn of a new day in cough science and therapy. Researchers, pharmaceutical leaders, clinicians, and thousands of coughers are now able to continuously monitor their coughs. At Hyfe, we are learning disease-specific cough patterns, studying individual cough triggers, and the relationship between subjective and objective perceptions of cough. Novel therapies are being developed to help coughers, both pharmacological and behavioral.
With the immense advantage of no side effects, digital behavior cough suppression programs have great potential in improving the quality of lives for millions of chronic coughers worldwide — on its own or in combination with soon-to-be FDA-approved molecules entering this previously stagnant market. We can look forward to the wide availability of individually tailored digital suppression therapy, all guided by the cough patterns and trends of each person seeking treatment.
In the place of a physician asking, “How is your cough?” and an unsure patient’s reply, “Um, it’s pretty bad doc,” acoustic AI gives us the tools to know the exact number and timing of coughs. It enables us to count the number and duration of bouts of multiple coughs. We can time the silences between coughs, when people with problematic cough get some respite. And we can precisely identify inflection points when cough patterns change, enabling prediction of COPD exacerbations, asthma attacks, and heading off emergency rooms trips before they are needed.
In low- and middle-income countries, community health workers, equipped with basic smartphones, will be able to help identify hard-to-find cases of tuberculosis, getting more people in treatment, increasing medication adherence, and reducing the spread of drug-resistant strains. Cough science will also improve our ability to diagnose children with malaria and pneumonia, two of the leading causes of childhood death globally.
The insights from cough monitoring will complement other streams of data to suggest best, evidence-based interventions promoting long term health: from prompting us to breathe through our nose to seamlessly controlling our indoor air quality through smart home devices, in cases of symptoms related to pollen, smoke, or smog. Infections will be detected earlier, even before we can notice the onset of symptoms. Management of chronic illness will perhaps benefit the most, by promoting and rewarding individualized, holistic interventions that improve quality of life.
Finally, syndromic acoustic surveillance will have immense value for public health systems around the world, spotting new pathogens and variants before they have spread. Acoustic AI will give us the ability to detect deviations from typical levels of cough quickly, allowing ultrafast, targeted screening and efficient prevention measures to be put in place in a timely manner.
This article was written by Mindaugas Galvosas, MD, Medical Officer and Business Development, and Reid Moorsmith, Chief Experience Officer at Hyfe AI, Wilmington, DE. Dr. Galvosas is pioneering the field of acoustic epidemiology, cough monitoring, and end-to-end digital cough products at Hyfe. He is previously the co-founder of digital health startup Aichom, which built software to support families of people with dementia. As CEO, Reid leads the app and wearables business. Reid translates consumer wants and needs into product and market-building opportunities. He has run operations for global organizations in places such as Liberia and Myanmar and is based in Vancouver, BC. For more information, visit here .