Kakao Brain Predicts Protein Structures with AI... Accelerating New Drug Development

AI Protein Structure Prediction Model 'Solvent' Released

Kakao Brain has introduced 'Solvent,' an artificial intelligence (AI)-based framework for predicting protein structures. It is expected to lower the research entry barriers for protein structure prediction and accelerate the pace of new drug development.

Kakao Brain's AI New Drug Research Team announced the release of Solvent on the 19th. Solvent is a model specialized for complex protein structures. It predicts protein folding (the process by which a protein's amino acid sequence folds into a three-dimensional structure) based on a single sequence. Its speed is at least three times faster than that of global companies' AI for protein structure prediction.

Kakao Brain has made the Solvent training code available on GitHub, a developer community. This is to help AI new drug development researchers carry out research tasks more easily and to contribute to the advancement of the AI new drug development ecosystem. Developers can use it to more easily develop or test customized models. Kakao Brain applied large-scale AI technology to Solvent to assist its use. Speed and memory efficiency have been improved by about 30% compared to existing training codes.

Kakao Brain is accelerating healthcare research using AI. A paper on chest X-ray language-image pre-training using large-scale AI was listed as a poster at 'MICCAI2023 (Medical Image Computing and Computer Assisted Intervention),' the top international conference in the medical image processing field. The paper dealt with research that efficiently matches X-ray images and medical reports using an already publicly available chest X-ray dataset. Kakao Brain demonstrated that AI technology can be usefully applied in the field of X-ray interpretation by producing results that efficiently increase disease detection rates compared to existing methods.

Additionally, Kakao Brain ranked third in the image captioning challenge held by Shutterstock, LG AI Research, and Seoul National University AI Graduate School during the world's largest computer vision conference, the 'Computer Vision and Pattern Recognition (CVPR) 2023.' Based on this achievement, they plan to enhance the performance of a draft reading report generation model for chest X-rays.

Kim Il-du, Co-CEO of Kakao Brain, said, "We are very pleased to achieve successful results in the AI healthcare sector, one of the company's main research areas," adding, "We will continue to research the AI healthcare field to improve human health through large-scale AI."

This content was produced with the assistance of AI translation services.