Japanese AI Startup Transforms Routine Throat Exams Into Rapid Disease Detection

A Tokyo-based startup is reimagining one of medicine's oldest practices. Iris, founded by former emergency physician Sho Okuyama, developed Nodoca, an AI system that analyzes throat images captured with a portable camera to detect influenza and other conditions in roughly 10 seconds. The approach sidesteps the conventional nasal swab test—notorious for its discomfort—and works even in the early stages of infection.

The system became Japan's first AI-powered medical device approved as a "new medical device" in 2022, and it's now installed in over 2,000 medical facilities across the country. In October 2025, it gained clearance to also assess COVID-19 infections. Okuyama's early career on remote islands, where he practiced medicine with only a stethoscope, shaped his conviction that better sensing and data acquisition—not just algorithmic brilliance—separate transformative medical AI from hype.

From Sparse Training Data to Competitive Moat#

When Iris launched in 2017, no AI training dataset focused on throat anatomy existed. Okuyama distributed cameras to roughly 100 clinics and spent three years collecting anonymized images with patient consent. This groundwork unlocked three critical challenges: building proprietary hardware, assembling training data, and proving the approach actually worked in practice.

Once commercialized, Nodoca's accuracy accelerated. The system ingests thousands of new annotated throat images daily in clinical use, growing its dataset from hundreds of thousands pre-launch to several million images today. This network effect acts as a moat—newer competitors entering the same space have no such accumulating advantage, and none have emerged to challenge Nodoca's position.

Broader Horizons Beyond Infection Detection#

The throat's visual patterns—its mucous membranes, blood vessel networks, and surface textures—encode more than signs of viral infection. Iris is researching whether AI can spot lifestyle-related diseases like diabetes and hypertension by detecting changes in these patterns. Okuyama describes a future where "a single photograph of the throat can be used to perform a comprehensive range of tests," with the cost of running additional AI inferences remaining negligible.

Redesigning Healthcare Systems#

Okuyama's ambition extends beyond a single diagnostic tool. He envisions AI seamlessly integrated across the entire care pathway—from intelligent intake interviews and remote consultations to in-person primary care and specialist referrals. To advance this vision, he has engaged actively in regulatory advocacy, pushing for policy reforms that align with what he sees as the logical next step in healthcare delivery.