講演情報
[P29-6]希少疾患に対するAI診断支援システムを用いたコンサルテーション体制の実践
○要 匡1, 五十嵐 ありさ1, 飯田 貴也1, 青木 大芽1, 柳 久美子1, 佐藤 万仁1, 成富 研二2, 松原 洋一3 (1.国立成育医療研究センター ゲノム医療研究部, 2.沖縄南部療育医療センター, 3.国立成育医療研究センター)
Rare diseases are epidemiologically estimated to affect approximately 5-8% of the overall population and 350 million people worldwide. However, the huge variety of diseases and the increasing their number year by year make diagnosis of individual cases often difficult. In Japan, comprehensive genome analysis is being conducted under the Initiative on Rare and Undiagnosed Disease (IRUD) and the Whole Genome Analysis Programmes, contributing to the diagnosis of rare diseases.However, at present, it is not cost-effective to perform a screening, such whole genomic analysis on all patients. It is important to consider cases well in advance, including listing appropriate differential diagnoses and analysing candidate genes, and such pre-screening is considered to increase overall efficiency.
As one of the supports for rare disease diagnosis, Prof. Naritomi constructed a genetic disease database (UR-DBMS) and created and published the Syndrome Finder (http://syndromefinder.ncchd.go.jp/) to support diagnosis. Based on the data from the UR-DBMS, we are developing a diagnosis support system using artificial intelligence (AI) with the aim of creating a more versatile system. The current system is a symptom-based diagnostic support system, with an accuracy of approximately 82% in studies of fixed diagnosis cases.
This time, using the AI diagnostic support system, the consultation-based operation was actually implemented. As a result, diagnostic efficiency increased, for example, genetic testing (gene-panel testing) of candidate diseases prior to entry into IRUD resulted the diagnosis.
As one of the supports for rare disease diagnosis, Prof. Naritomi constructed a genetic disease database (UR-DBMS) and created and published the Syndrome Finder (http://syndromefinder.ncchd.go.jp/) to support diagnosis. Based on the data from the UR-DBMS, we are developing a diagnosis support system using artificial intelligence (AI) with the aim of creating a more versatile system. The current system is a symptom-based diagnostic support system, with an accuracy of approximately 82% in studies of fixed diagnosis cases.
This time, using the AI diagnostic support system, the consultation-based operation was actually implemented. As a result, diagnostic efficiency increased, for example, genetic testing (gene-panel testing) of candidate diseases prior to entry into IRUD resulted the diagnosis.