As the lead product designer, I was responsible for the end-to-end design of the mobile app. I worked with a multidisciplinary team of engineers, pulmonologists, and researchers across four countries to conduct user research, design the computer vision guidance system, create a trustworthy results interface, and establish a cross-platform design system.

70% of the 300 million people with COPD go undiagnosed, largely because spirometry is slow, expensive, and requires specialists, leaving “diagnostic deserts” worldwide.

Eupnoos built breakthrough AI to analyze breath sounds, but the real challenge was earning doctors’ trust: convincing them to choose software over $20,000 hardware and making the AI transparent in a field where trust is critical.

Throughout the whole duration of the project, I took feedback from healthcare providers across multiple countries, from pulmonologists in UK hospitals to maritime health coordinators in Singapore.

Healthcare providers weren't resistant to smartphone diagnostics because of conservatism - they were traumatized by previous failed implementations. I learned that ~60% of new medical technologies are abandoned within six months because they don't integrate smoothly with existing workflows.

I documented user flows across vastly different environments: Singapore shipyards with industrial noise, UK hospitals with controlled lighting, and rural clinics with limited connectivity.

The existing respiratory screening workflow was shockingly manual. Traditional spirometry required specialized rooms and trained technicians. Our interface needed to collapse this 45-minute process into something achievable anywhere, in minutes.

The core technical challenge was ensuring consistent breath capture without a physical device. To solve this, we replaced the physical constraints of a traditional spirometer with digital constraints.

I designed a real-time positioning system that uses the phone's camera to act as a "magic mirror." The interface uses spatial overlays to guide the user on optimal phone-to-face distance and provides instant, clear feedback like "TOO LOW" to correct positioning. This ensures standardized, repeatable test quality, mimicking the coaching of a respiratory therapist.  

I created a results presentation that speaks healthcare providers' language rather than hiding behind simplified consumer interfaces. The results screen displays traditional medical metrics including FEV1A (Forced Expiratory Volume), FVC A (Forced Vital Capacity), and PEF A (Peak Expiratory Flow) with actual clinical values that match spirometry standards.

Clear diagnostic categories show Asthma, COPD, and Lung Health status with recognizable medical terminology. Visual data representation uses familiar spirometry curves that healthcare providers already understand.

My focus was evolutionary improvement over flashy redesigns, building trust through familiarity.

The Singapore maritime pilot with SembCorp Marine presented unique challenges. I designed for 700+ shipyard workers with high humidity affecting touchscreens, industrial noise interfering with recording, and bright sunlight making screens unreadable.

My design adaptations included high contrast visual elements optimized for sunlight, large touch targets for gloved hands, noise-resistant audio guidance with visual backup instructions, and simplified workflow logic.

The interface enabled mass respiratory screening processing, reduced screening time from 45 minutes to 5 minutes while maintaining diagnostic accuracy, and provided early COPD detection for previously undiagnosed rural patients.

Eupnoos demonstrated how thoughtful interface design could democratize access to advanced medical diagnostics, making sophisticated AI-powered healthcare tools available where traditional medical equipment would be impossible to deploy.

Designing for a life-critical application like Eupnoos taught me that the primary job is to build trust. This isn't achieved with revolutionary interfaces, but by making sophisticated technology feel reliable and transparent. Clinicians need to understand how an AI reaches its conclusions, not just see the results.

The process also proved that designing for extreme environments, like a noisy shipyard, forces you to build a more accessible solution for everyone, and that the best medical interfaces emerge from a deep collaboration between engineering, clinical expertise, and design.