A routine throat check may look simple, but Iris believes it contains more useful medical information than conventional exams have been able to capture. The Japanese company’s Nodoca system uses artificial intelligence to analyze images of the pharynx, combining those images with information from the patient’s medical interview and other data to return an influenza assessment in a little over 10 seconds.
The appeal is practical. Instead of relying on a conventional influenza test that requires a swab to be inserted deep into the nose to collect a nasal specimen, Nodoca uses a small camera-equipped device to examine the throat. According to the source article, it can also be used at an earlier stage after symptom onset.
Why the throat matters
Throat exams are among the most basic parts of a physical exam, sitting alongside familiar checks such as body temperature and heartbeat. Doctors have long looked into the throat because visible findings can offer clues about what is happening in the body.
The core insight behind Nodoca is that those findings may be richer than a clinician can easily interpret in real time. Different pathogens can create different patterns in the throat. Iris is trying to make those patterns more readable by pairing a dedicated imaging device with AI analysis.
That focus matters because the system is not only about software making a decision. Sho Okuyama, the leader of Iris and a former emergency physician, emphasizes the importance of capturing the right data in the first place.
“When people hear ‘medical AI,’ they tend to think of the part that makes diagnoses,” Okuyama says. “But the real differentiating factor lies in sensing—how the data is acquired. We developed our own hardware and ventured into the area of pharyngeal images, which no one had touched.”
In plain terms, Iris had to build both the way to see the throat and the way to interpret what was seen. That made the project more difficult than applying an existing algorithm to an established dataset.
From scarce data to clinical use
When Iris was founded in 2017, there was no training data for AI focused on the throat. Okuyama addressed that gap by lending dedicated cameras to around 100 medical institutions. Over about three years, the company collected data with patients’ consent.
That early work had to answer three basic uncertainties: whether the hardware could be developed, whether enough training data could be gathered, and whether AI-assisted diagnosis could be made practical. The company’s progress on those fronts led to commercial implementation.
In 2022, Nodoca became the first AI-equipped medical device in Japan to receive approval as a “new medical device” and to gain coverage under the national health insurance system. Since then, it has been introduced at more than 2,000 medical institutions across Japan.
The system has continued to expand. In October 2025, it received approval for an additional function related to Covid-19. The source article does not describe the function in detail, but the approval shows that Nodoca is not being treated only as a single-purpose influenza tool.
How use makes the system stronger
Once Nodoca entered clinical practice, its data advantage began to grow. Each use can add anonymized processed data to the system. Before launch, the company had hundreds of thousands of throat images; that figure has now grown to several million.
This creates a network effect. More clinical use means more data, and more data can support improving accuracy. The source article describes that accumulation as a barrier to entry for competitors and notes that no latecomers have emerged in the same field.
The practical value is easy to understand. A tool that can assess influenza from pharynx images in just over 10 seconds could make one uncomfortable part of routine care less burdensome. It also puts a familiar exam, the throat check, into a more data-driven workflow.
- Device: a compact camera-equipped system for pharynx images.
- Company: Iris, led by Sho Okuyama.
- Current role: AI-assisted influenza assessment.
- Clinical reach: more than 2,000 medical institutions across Japan.
- Data scale: throat images have grown from hundreds of thousands before launch to several million.
Beyond infectious disease
Iris is also looking past influenza and Covid-19. The company is conducting research and development on AI that may detect lifestyle-related diseases such as diabetes and hypertension by identifying changes in patterns in the throat’s mucosa and blood vessels.
That is still described as research and development, not as an approved everyday use. But it explains why Iris views the throat as more than a place to check for infection. If a single image can reveal multiple patterns, then one capture could eventually support a wider range of tests.
“I think that in about 10 years, we'll reach an era in which a single photograph of the throat can be used to perform a comprehensive range of tests,” Okuyama says. “The cost of AI inference is extremely low, so even if we increase the number of tests, the cost of providing them hardly changes.”
The logic is straightforward: once the image is captured, additional AI analysis may not require a comparable increase in cost. The source does not claim that this future has arrived, but it frames the direction Iris is pursuing.
A wider AI health care vision
Okuyama’s ambitions extend beyond one device. He is considering whether AI can become a foundation for redesigning health care itself. The source article describes a path that includes AI-powered medical interviews, online consultations, in-person care from primary care physicians, and treatment at specialist hospitals.
In that vision, AI would appear at multiple stages rather than only at the moment of diagnosis. Okuyama has also been proposing regulatory reform and lobbying government ministries as part of the effort to make that redesign possible.
“The ‘future of health care’ that I talk about doesn't actually include a single future technology,” he says. “Everything can be achieved with technology that already exists today. Our job is to close the gap that makes these things have to be talked about as ‘the future.’”
Iris is also preparing for overseas expansion. Because the body’s mucous membranes are believed to be highly consistent, Okuyama says the data and algorithms accumulated in Japan are expected to be transferable elsewhere. Clinical trials in the United States are also under consideration.
For now, Nodoca’s significance is concrete: it turns a basic throat exam into an AI-assisted influenza assessment that can avoid the conventional nasal swab. Its larger implication is just as clear. Some of the most useful medical AI may begin not with exotic new tests, but with better ways to read ordinary signals the body already shows.