Why We Built an AI Tool to Analyze Coughs

Somewhere between a scratchy throat and a full-blown doctor's visit, there's this frustrating gray zone most of us know all too well. You've got a cough. It's been going on for a week, maybe two. Is it just a cold working its way out? Allergies acting up again? Or is it something that actually needs a professional to look at? Most people don't have a great way to answer that question — so they either ignore it longer than they should, or they book an appointment for something that turns out to be nothing.

That gray zone is exactly why we built Know Your Lung Health, or Kylunghealth for short. This post is our attempt to explain, in plain language, the story and reasoning behind our AI cough analysis tool — why we thought it was worth building, what problem it's actually trying to solve, and where we hope it goes from here.

The Problem We Kept Running Into

Cough is one of the most common symptoms in all of medicine. It's also one of the most frustratingly vague. Two people can describe "a bad cough" and mean completely different things — one might have a dry, tickly cough from seasonal allergies, another might have a chesty, productive cough tied to early bronchitis, and a third might have a persistent cough that's been quietly building for weeks because of undiagnosed asthma.

The traditional way to sort this out is simple in theory: see a doctor, get examined, maybe get listened to with a stethoscope, and go from there. But in practice, that path has real friction. Appointments take time to book. Clinics are busy. Many people — especially those managing mild or moderate symptoms — end up putting it off simply because it doesn't feel "serious enough" to justify the effort. And by the time some coughs do get checked out, they've already turned into a chronic cough that's been lingering for a month or more.

We kept seeing this pattern over and over: people living with a cough that clearly wasn't going away, unsure if it warranted medical attention, with no easy way to get even a preliminary sense of what might be going on. That gap — between "I have a symptom" and "I know whether this needs attention" — is what got us thinking seriously about AI cough analysis as a potential bridge.

Why a Cough, Specifically?

It might seem like an odd starting point for a health-tech company. Why not blood pressure, or sleep, or something more commonly tracked? The answer comes down to how much information is actually packed into the sound of a cough.

A cough isn't just noise — it's your respiratory system's forceful attempt to clear something out, and the way it does that varies meaningfully depending on what's happening inside your airways. A dry cough sounds and behaves differently from a wet cough. The acoustic pattern of a cough linked to asthma differs from one linked to a simple cold. Clinicians have long used the sound and pattern of a cough — alongside other symptoms — as a diagnostic clue during physical exams. We asked a simple question: if trained ears can pick up on these patterns, could a properly trained AI model learn to recognize similar acoustic signals, consistently, at scale, and make that insight accessible to more people?

That question became the foundation of Kylunghealth's AI cough analysis.

Building Something People Could Actually Use

From the start, we didn't want to build something that only made sense inside a hospital or research lab. We wanted a tool that anyone dealing with a persistent cough, a wheeze they couldn't explain, or a general sense that something felt "off" with their breathing could use immediately — from their phone, without needing an appointment first.

That meant a few things had to be true about the product:

It had to be fast. Nobody wants to sit through a 20-minute intake process just to get a basic read on a cough. Our assessment takes just a couple of minutes.

It had to be simple. You record a short cough sample, and our AI analyzes the sound characteristics — no technical knowledge required to understand your results.

It had to be honest. This is genuinely important to us, and we say it clearly throughout our platform: Kylunghealth is a screening support tool, not a diagnostic device. We're not trying to replace your doctor. We're trying to give you a useful starting point before you decide whether a doctor visit is the right next step.

That last point matters more than it might seem. Health tech has a track record of overpromising, and we didn't want to add to that. What we can offer is a fast, accessible respiratory risk overview based on AI cough analysis — a tool that helps close the gap between "I have a symptom" and "I know what to do about it," without pretending to be something it's not.

What Our AI Actually Listens For

At a high level, our system analyzes the acoustic patterns in a recorded cough — things like frequency, duration, and other sound characteristics that have been associated with different respiratory presentations. This is conceptually similar to how a clinician trains their ear to notice the difference between a dry, tickly cough and a wet cough with phlegm, or to pick up on subtle patterns tied to airway narrowing.

We're intentionally cautious about how we describe this. The World Health Organization has highlighted the growing role of digital health tools in expanding access to care, particularly in settings where traditional screening is harder to access — and we see AI-based respiratory screening as part of that broader shift. But we also know that meaningful clinical decisions require a full picture: history, physical examination, and appropriate testing, not a single data point from an app. Our tool is designed to complement that process, not substitute for it.

The Bigger Picture: Respiratory Health Shouldn't Feel Like a Mystery

One thing we kept coming back to while building Kylunghealth is how little most people actually know about their own respiratory health. We can tell you our resting heart rate, our step count, how many hours we slept last night — thanks to how normalized fitness tracking has become. But ask most people what a wheeze actually means, or why a persistent cough matters more after three weeks than after three days, and you'll usually get a shrug.

We think that gap in understanding is a real problem, and it's one of the reasons our platform goes beyond just the AI cough analysis tool itself. We've built out educational content covering everything from the difference between dry cough and wet cough, to what wheezing and crackles actually mean, to how conditions like asthma and COPD tend to show up in everyday coughing patterns. Our goal is for Know Your Lung Health to genuinely live up to its name — helping people understand their lung health, not just screen it.

You can explore more about how the underlying technology works on our How It Works page, which breaks down the process in plain language without the technical jargon that so often makes health tech feel inaccessible.

Read More - Respiratory Health Screening: How AI Technology Helps You Understand Your Lungs

Who This Is Actually For

We built Kylunghealth with a few different people in mind, and if you see yourself in any of these, there's a decent chance our tool can help:

• The person who's had a cough for three weeks and keeps putting off calling the doctor. A quick assessment can help you decide whether it's time to actually make that call.
• The parent trying to figure out if their kid's cough sounds "normal" or worth a pediatrician visit. Understanding common patterns — even just reading through our educational content — can bring a little more clarity to a stressful moment.
• Anyone managing a chronic respiratory condition who wants an easy way to keep an eye on changes in their cough over time, between regular checkups.
• The naturally curious person who just wants to understand their own body better, the same way they might track sleep or activity levels.

We're also proud that our platform has found real relevance with healthcare providers, who use Kylunghealth as a pre-visit screening aid and a way to support remote monitoring for patients managing ongoing respiratory conditions. You can learn more about that side of what we do on our page for healthcare providers.

What's Next

We're not done. Cough analysis is our starting point, not our finish line. Our team is actively working on expanding what Kylunghealth can detect and support, including additional respiratory sound modalities beyond cough, and deeper condition-specific insights — developed carefully and in coordination with planned clinical research rather than rushed to market ahead of proper validation.

We think that's the responsible way to build health technology: start with something genuinely useful and honestly described, keep listening to how people actually use it, and expand capabilities as the science and data catch up with the ambition. Organizations like the Mayo Clinic have long emphasized that respiratory symptoms like cough, wheezing, and shortness of breath deserve careful, individualized evaluation — and that's a principle we've tried to build into Kylunghealth from day one, even as we push to make that evaluation process more accessible.

Lessons We Learned Along the Way

Building Kylunghealth wasn't a straight line, and we think it's worth being honest about that too. Early on, we underestimated just how much variation exists in something as seemingly simple as a cough. A dry cough from one person can sound remarkably different from another person's dry cough, depending on age, underlying health, even the microphone quality of the device they're using. Getting an AI system to recognize meaningful patterns across that kind of variability — rather than just memorizing a narrow set of examples — turned out to be one of the harder engineering challenges we faced.

We also learned, sometimes the hard way, how important it is to be conservative with our claims. It would have been easy to market Kylunghealth as something that "diagnoses" conditions, because that language sounds more impressive and might generate more attention. But we made a deliberate choice not to go down that road. Overstating what a tool like this can do doesn't just risk misleading people — in a health context, it can actually cause harm, whether that's false reassurance or unnecessary anxiety. We'd rather be precise and trustworthy than flashy and misleading, even if it means our messaging sounds more measured than some competitors in the digital health space.

Another thing we didn't fully anticipate was how much people wanted to understand their symptoms, not just get a number or a score. That's part of why our platform leans so heavily into educational content alongside the actual assessment tool. Knowing that your cough pattern falls into a certain risk category means a lot more when you also understand what a persistent dry cough might indicate, or why wheezing and shortness of breath together are worth paying closer attention to. Data without context isn't nearly as useful as data paired with understanding — and we've tried to build Kylunghealth around that principle.

The Human Side of a "Tech" Product

It's easy to talk about AI cough analysis in purely technical terms — algorithms, acoustic features, model training — and lose sight of what actually matters: real people dealing with a real symptom that's disrupting their sleep, their work, or just their peace of mind. Every design decision we've made, from how simple the recording process is to how plainly we explain results, comes back to that same question: is this actually helpful for someone who's a little worried about their cough and doesn't know what to do next?

We've heard from users who put off seeing a doctor for weeks because their persistent cough didn't feel "bad enough" to justify an appointment — until a quick assessment gave them enough clarity to finally make that call. We've heard from parents trying to understand whether their child's nighttime coughing was something to flag to a pediatrician. These are the moments that remind us why we started building this in the first place, and they're a big part of why we keep investing in making Kylunghealth more accurate, more accessible, and more genuinely useful over time.

Why It Matters to Us

At the end of the day, we built this AI cough analysis tool because we genuinely believe people deserve an easier, faster, less intimidating way to start understanding their respiratory health. Not a replacement for medical care — a bridge toward it. A way to move from uncertainty to a little more clarity, in the time it takes to record a short cough sample.

If you've been sitting with a cough you're not sure about, wondering if it's worth mentioning to a doctor, that uncertainty is exactly the gap we set out to close.

Curious what your cough might be telling you? Try Kylunghealth's free AI cough analysis and take the first step toward understanding your lung health.