Silicon Valley giant NVIDIA is teaming up with pharma company AstraZeneca and the University of Florida on new artificial intelligence research projects aimed at boosting drug discovery and patient care.

This morning NVIDIA and AstraZeneca revealed a new drug-discovery model called MegaMoIBART, which is aimed at "reaction prediction, molecular optimization and de novo molecular generation." MegaMoIBART will be deployable on NVIDIA's platform for computational drug discovery, known as Clara Discovery, and will use a new kind of technology called transformer neural networks.

"They've only been available in the last couple of years. What is important to understand about them is it’s a two-stage process," said Kimberly Powell, VP of healthcare at NVIDIA. "The first stage is an unsupervised learning stage where you are pre-training a very large model with unlabeled data. Then it is followed by a specific task fine-tuning training with smaller amounts of labeled data to allow a researcher to train on enormous unlabeled data sets and then fine-tune for much more task-specific problem sets."

According to Powell, transformer neural networks could help accelerate research and speed up drug discovery.

"A training method like this could avoid the need for very large labeled data sets. In this digital biology revolution… these data sets cannot be labeled. But there is so much information we can extract from them that no human could, so these transformer models are absolutely necessary and they can really help us address some challenges in drug discovery and healthcare overall," according to Powell.  

The platform is being trained specifically on a chemical compound data set using NVIDIA DGX SuperPOD. 

Meanwhile, the partnership with Florida is focusing on a new model called GatorTron, which was developed to match patients to clinical trials, as well as, predicting life-threatening conditions and providing clinical decision support to doctors.

According to NVIDIA, the platform can also help boost drug discovery by creating "patient cohorts for clinical trials and for studying the effect of a certain drug, treatment or vaccine."

WHY IT MATTERS

Drug research and discovery is expensive and time-consuming. According to a study published in Nature, the overall failure rate for drug development is more than 96%, and the failure rate during clinical development is 90%. However, pharma companies and tech companies have been increasingly turning to AI to tackle some of these issues.

"We are digitizing biology at a pace we never have before that is going to infuse a new way of doing drug discovery. We need to have new approaches with AI because that level of data – that amount of data is too much for any human to be able to understand," Powell said. "So, AI approaches in drug discovery are really going to create that next generation of computational drug discovery."

THE LARGER TREND

While NVIDIA may be classically associated with the video game industry, it has been working in the health sector for some time. In 2018 the company rolled out its AI platform Clara, which was designed to use AI to create a virtual medical imaging platform. Since then the company has launched a Clara AI toolkit for radiologists.

The company has also been active in the genomics space. In March NVIDIA announced its plans to team up with Harvard University on a new AI-based toolkit designed to help researchers gain more access and insights into DNA. Additionally, the tech giant announced it would build a supercomputer for U.K. healthcare researchers.

During the pandemic the company also demonstrated an interest in telehealth, rolling out its automated speech recognition and natural language processing technology that can transcribe and organize information from a telemedicine visit for patients and clinicians.