Seoul National University Hospital Consortium Initiates Project to Build Super Large AI Data for Children's Abdominals

Jul 28, 2024

The Seoul National University Hospital consortium announced that it will carry out a project to build 'Children's Abdominal Multimodal and Synthetic Data' to spread ultra-large AI and respond to on-site demand. Through this data construction project, it is expected that the health care and treatment results of pediatric patients will improve as the accuracy of diagnosis of pediatric abdominal diseases improves.

This project is part of the 「Early 2024 AI Spread Ecosystem Creation Project」 organized by the Ministry of Science and ICT and promoted by the Korea Intelligence Information Society Promotion Agency (hereinafter referred to as"NIA"). The Seoul National University Hospital Consortium was selected as the final implementation institution for the construction project of the health care sector 'Children's Abdominal Multimodal and Synthetic Data'.

Seven organizations, including ▲ Gyeongbuk University Industry-Academic Cooperation Group ▲ Korea University Industry-Academic Cooperation Group ▲ Gil Medical Foundation ▲ Yangsan Busan University Hospital ▲ Urban Data Lab ▲ Sur, are participating in the consortium, with a budget of 1.2 billion won.

A report on the start of the project was held with the NIA at Seoul National University Children's Hospital on the 11th. At the meeting, the Seoul National University Hospital consortium emphasized the importance of enhancing competitiveness in AI services and building high-quality and large-scale data at the national level. In particular, the importance of establishing data to improve treatment outcomes for childhood abdominal diseases is being highlighted at a time when access to childhood-related imaging images is gradually decreasing due to a decrease in fertility rates.

The data constructed in this project consists of x-ray and medical imaging image data for the diagnosis of pediatric abdominal disease. The data is divided into two parts. Multimodal data include approximately 2000 pairs of x-rays and other medical imaging and clinical data from the same patient. Based on actual multimodal data, about 10,000 synthetic data will be constructed from data that have been labeled for major clinical symptoms, diagnosis names, and treatment methods. These data will be used to develop artificial intelligence models that assist in diagnosing childhood abdominal diseases.

The Seoul National University Hospital consortium plans to create a more accurate model by collecting multimodal data and combining various information. Based on this, it is expected to increase data diversity and solve data shortages by generating synthetic data through labeling work and artificial intelligence techniques. The project will run until the end of this year and will be unveiled through 'AI-Hub' operated by NIA after the construction is completed.

Professor Kim Hyun-young (Children's Surgery) of Seoul National University Hospital said as the general manager of this project, `The ultra-large AI data construction project will be a new leap forward using AI technology in the diagnosis and treatment of childhood abdominal diseases.' `We expect this project to greatly improve the accuracy and efficiency of diagnosing childhood abdominal diseases.'



Seoul National University Hospital Consortium Initiates Project to Build Super Large AI Data for Children's Abdominals
Kim Sung-hyun, chief of the National Intelligence Service (left) and Professor Kim Hyun-young of Seoul National University Hospital


bellho@sportschosun.com