Cough analysis tool screens for TB via smartphone

This exclusive interview with Venkat Yechuri, CEO of Salcit Technologies, explores how AI is transforming TB screening in underserved communities. This is part one of a two-part series. In this first installment, Yechuri details Salcit’s innovative cough analysis technology and vision for expanding beyond TB diagnosis. Read part two for insights on Salcit’s strategic partnership with Google and expansion plans.

Salcit Technologies was founded to address a critical healthcare challenge: 50% of outpatients in India’s healthcare system report having some kind of respiratory issue, but there’s no way to diagnose them, especially in rural areas with no access to the expensive equipment and trained technicians needed for testing.

Salcit is helping to address this challenge with AI-powered cough analysis that can be conducted without specialized training using only a smartphone. Salcit’s team has spent years collecting the world’s largest physician-annotated cough database, a repository used to train algorithms developed by the company to help diagnose tuberculosis.

For the past five years, Venkat Yechuri has served as Salcit’s CEO. With a professional background in engineering and manufacturing, Yechuri has leveraged his expertise in scaling production to a global level to help accelerate the adoption of Salcit’s SaMD solution, Swaasa®, to regions beyond India.

Yechuri sat down with DHI to discuss how smartphone-based AI diagnostics are expanding access to TB screening, the challenges of scaling medical technology across rural regions, and the potential for acoustic biomarkers to revolutionize remote patient monitoring. The conversation was conducted over two parts, with the second installment exploring how recent advances in AI are enhancing the technology’s capabilities in the field.

Why did Salcit begin with a focus on helping diagnose tuberculosis?

Yechuri: The UN’s sustainable development goals include eliminating TB by 2030, but TB is a hard nut to crack—screening for it brings a number of challenges, especially in specific localities. For example, in Indonesia there are thousands of remote islands, and a very high rate of TB. Medical science has figured out how to treat TB once it’s found, but how do you find it in that situation?

Swaasa is helping us reach the UN’s goal by making it possible for community health workers in rural areas to diagnose TB simply by recording a patient’s cough sounds for 10 seconds. Swaasa “listens” to the recording and shows a red, green, or amber color—if a red is displayed, the patient can be referred immediately for TB treatment.

That’s how simple Swaasa is to use, and why we started with TB. We’ve run clinical validations in India in the past independently with funding from the U.K. Government, and are running another one now. We are also working with the University of California, San Francisco, and are operating in five countries.

Will Swaasa eventually be able to help diagnose respiratory conditions other than TB?

Yechuri: We are currently exploring the possibility of taking our understanding of cough and applying it to other conditions, both pediatric and adult, and seeing if it is possible to diagnose if there is an issue. For example, if a community health worker screens a patient with our product and sees the red light, they know the next step is to refer the patient to a nearby physician. While a physician might be better equipped to make a diagnosis, they may not be able to reach a differential diagnosis identifying if the patient has asthma, COPD, etc.

At this point in our development of Swaasa, we do not make any diagnostic claims. But our aim is to make it possible for Swaasa to help point physicians in the right direction with their diagnosis. In the context of countries like India, where the patient-to-physician ratio is so high and doctors are hard-pressed to give a patient more than two minutes of examination time, this assistance from Swaasa could help enable better patient outcomes.

Do you have plans to expand Swaasa’s capabilities beyond cough analysis?

Yechuri: Our roadmap includes making it possible for Swaasa to analyze acoustic biomarkers other than cough while keeping it usable on any smartphone. One of these biomarkers is speech—speech analysis through Swaasa could be potentially helpful in diagnosing conditions beyond respiratory ones, including those related to heart and mental health.

Along with speech, all the acoustic biomarkers on our technology roadmap will likely take at least three years of work before they lead to a product release, including data gathering and validation, clinical validation, and publication. It’s a long journey.

Are there other areas of patient care in which you see Swaasa having an impact?

Yechuri: In time, I believe Swaasa may also be able to help with post-diagnosis monitoring. Once a physician has diagnosed a patient with a respiratory disease such as COPD, there is no equivalent to the home monitoring that exists for conditions involving blood glucose or blood pressure. There are “peak flow meters” and handheld spirometers, but physicians cannot rely on their results because they require patients to breathe in a specific manner that can be subjective.

Swaasa can take the subjectivity out of the equation by simply asking a patient to cough for 10 seconds and analyzing the recording to tell if it’s a “good” cough sample or a “bad” cough sample. If it’s a good enough sample, Swaasa can then analyze it and provide the results. This would be a huge advancement, especially in the United States within the rapidly growing area of remote patient monitoring.

If you’ve got patients you’re managing remotely, Swaasa can potentially provide you with reliable data. You can see what a patient’s baseline was six months ago and how they are trending, and check this as many times as needed because it only requires a patient to cough into a smartphone. To aid physicians with monitoring this data, Swaasa supports the export of data into standard CSV format, and our developers are working on APIs for EHR integration. Data export to FHIR formats is planned for the near future, as well.

We have more to accomplish on the monitoring side, but the potential to improve patient outcomes and relieve hospital expenses is massive. Just take COPD—it’s a leading cause of death in the U.S. and a $50 billion problem a year, with $15 billion coming from patients ending up in the emergency room because of exacerbations that could be avoided if we could get these patients to a physician or nurse sooner.

That’s one of our biggest goals with Swaasa, the accurate prediction weeks in advance that a patient is heading for an exacerbation in their condition and helping to facilitate proactive intervention.

This interview has been edited for brevity. Read part two here.