NVIDIA today announced that the new AI Center at Mahidol University (MU) in Thailand has adopted NVIDIA DGX A100 systems and the NVIDIA Clara platform for healthcare to power drug discovery research, with a focus on genomics, proteomics, pathology and radiology.
Housed at the university’s Salaya campus, the center will enable multi-disciplinary research collaboration and allow researchers to solve the world’s most challenging problems on the NVIDIA-based platform using both hardware and software.
It will initially support projects from the Cluster of Excellence in AI-Based Medical Diagnosis, the Integrative Computational BioScience Center and the Artificial Intelligence-Integrated Drug Discovery Platform, from the Faculty of Information and Communication Technology, Faculty of Medicine Siriraj Hospital and Faculty of Pharmacy, respectively.
“We are targeting to have the platform support the end-to-end drug discovery process, using NVIDIA Clara Discovery for drug development, NVIDIA Clara Parabrickes for genomic analysis and Clara Imaging with ePAD and MONAI for radiology and pathology to innovate and accelerate our AI model creation and deployment,” said Dr. Pattanasak Mongkolwat, Dean of the Faculty of ICT, at Mahidol University, the oldest and one of the most prestigious universities in Thailand.
Designed to handle all AI workloads, NVIDIA DGX A100 enables organizations to consolidate training, inference and analytics into a unified, easy-to-deploy AI infrastructure. NVIDIA Clara Discovery is a collection of frameworks and AI models for in silico drug development for virtual screen, protein visualization, computational chemistry, docking, generative structure creation, protein function predictions, and natural language processing. NVIDIA Clara Parabricks is a GPU accelerated genomic analysis platform for next generation sequencing DNA and RNA data, capable of analyzing a 30x whole genome in 25 minutes versus 30 hours on a CPU-based bioinformatics pipeline. NVIDIA Clara Imaging is an application framework that accelerates the development and deployment of AI models in radiology and pathology with tools like AI-Assisted annotation tools for faster labeling of data, hyperparameter tuning, federated learning for collaborative robust model building across institutions, and over 20 pre-trained models. The framework has upgraded its underlying infrastructure from TensorFlow to PyToch-based MONAI, which is an open-source accelerated framework.
Bring Thailand’s University Community Together
The Mahidol University AI Center features state-of-the-art training and meeting rooms within reach of researchers in and outside of the university. It provides a conducive environment where researchers can collaborate and initiative new ideas and projects, and where MU students and researchers across disciplines can be trained on AI.
The center plans to bring researchers in Thailand’s university community together to work on disciplines such as computer science, engineering, life science, medicine, music for health, and social science and humanities.
Building on the university’s strengths in medicine and health sciences, it will further scientific knowledge and breakthroughs to support the United Nations’ sustainable development goals to make the world a better place.
As part of the Thailand AI University Consortium, the center will collaborate on developing an automated framework for AI assisted annotation and AI federated learning in medical imaging solutions. This will allow an image annotation workstation to be integrated into clinical workflows to reduce disruption during annotation for deep learning purposes, while maintaining data integrity, data security and data anonymity.
The federated learning platform can be expanded and linked to partner institutions. AI models and training results can be shared without being hindered by institutional data sharing policies. Anonymization mechanisms will be in place to automatically remove patient identifiable information and collect annotation data as structured data. Image data, along with corresponding annotation data, can be fed to a deep learning model for training and classifying.
“Each NVIDIA DGX A100 delivers five petaflops of AI power, offering unprecedented compute density, performance and flexibility to organizations. With the systems power Mahidol University AI Center, researchers in the university and across Thailand will be able to accelerate research breakthroughs that benefit Thailand and the world,” said Dennis Ang, director of enterprise business for the SEA and ANZ Region at NVIDIA.
Liked this post? Follow SwirlingOverCoffee on Facebook, YouTube, and Instagram.
