Rare diseases are estimated to account for about 5-8% of the population in total, but because more than 9,500 diseases exist and many prove difficult to diagnose that it can take years, even decades before a diagnosis is made, this result in an uncertain future for the patient, family, and attending physician.

In order to help such patients to move on to the next step, such as confirming the diagnosis as soon as possible, providing medical care and appropriate treatment for the disease, etc., we are building a “Rare Disease AI”. There are two main methods of AI-based diagnostic support for rare diseases: one is to estimate the disease based on facial and other images (image-based), and the other is to estimate the disease based on symptoms, findings, and other information (symptom-based) (Figure).

For the first step, the Center has first built a symptom-based AI diagnosis support system and has begun its utilization/consultation. In the future, we plan to combine and develop both methods to create a more flexible system.