AI Rare Disease Diagnosis System RareLens Launches in Japan: 94% Accuracy in Identifying Rare Diseases from Genomic Data

On July 1, 2029, Japan's Ministry of Health, Labour and Welfare announced the nationwide launch of RareLens, an AI rare disease diagnosis system, at 42 designated hospitals. The system analyzes patients' whole genome sequencing (WGS) data to provide rare disease diagnostic suggestions within 3 weeks, achieving 94% accuracy.

Globally, approximately 7,000 known rare diseases affect about 300 million people. Rare disease patients see an average of 7.3 doctors over 5.4 years before receiving a diagnosis. RareLens was jointly developed by RIKEN and Fujitsu, with training data covering genomic characteristics and clinical phenotypes of 4,800 rare diseases.

The workflow: physicians submit patient WGS data and clinical symptom descriptions; RareLens completes variant pathogenicity analysis, phenotype-genotype matching, and literature evidence retrieval within 48 hours, outputting a report with three most likely diagnoses. In retrospective testing involving 2,100 diagnosed patients, the correct diagnosis appeared in the top three candidates 94% of the time and ranked first 82% of the time.

Japan rare disease patient organization leader Emi Sato said: "For rare disease patients, diagnosis itself is the beginning of treatment. Many families experience years of misdiagnosis and anxiety before getting answers."