Enhancing Life Science R&D and Improving Data Accessibility with Vyasa and NVIDIA
COVID-19 has proved to be a forcing function for life science companies to improve processes to expedite the development and distribution of vaccines and other therapeutics. As a result, the life sciences industry is increasing its interest in solutions that can enhance drug discovery capabilities. For example, Deloitte found that 81% of large pharmaceutical companies are prioritizing investments in AI to address these needs.
State of Life Science R&D
The drug development process takes an average of ten years from discovery to approval, with the discovery and preclinical stages taking three to four of those years. During this time, life science companies spend an average of $35M on R&D per active trial, with that number reaching $66M per active Phase III trial. (1)
Unfortunately, many of these trials do not reach completion, with 19% of registered trials being terminated or failing to meet accrual goals. But what if terminated or failing trials could be identified earlier in the process, enabling life science companies to save critical time and financial resources?
R&D’s Data Challenge
One of the biggest hindrances to early research and preclinical stages is access to relevant and accurate data. This is because a growing amount of organizational data is siloed across departments and stored in unstructured formats, making it difficult to search and pull insights from. For example, only 40% percent of eligibility data comes from structured content.
This scenario requires teams to spend weeks or months manually searching through complex documents such as clinical trial protocols, scientific literature, images and more to identify the data they need to design a trial and move forward in the R&D process. In most cases, these teams only gain access to a fraction of the data they need, meaning insights are missed that could influence the efficacy and overall success of the trial.
Advanced Data Accessibility with an Intelligent Data Fabric
Vyasa is addressing this challenge head-on for life science companies. With Vyasa Layar, life science companies are able to connect disparate data silos, regardless of file structure or storage location, into an intelligent data fabric. Once data is connected, Layar applies deep learning on the content, making it easily searchable and accessible in low-code applications such as smart spreadsheets and dynamic knowledge graphs.
Vyasa Layar leverages the NVIDIA BioMegatron model within its GPU-accelerated data fabric architecture to provide solutions for biomedical harmonization, multi-institutional cancer research and clinical trial protocol design. NVIDIA’s NeMo toolkit is instrumental in building state-of-the-art conversational AI models, while the NVIDIA Triton Inference Server, which is included in the NVIDIA AI Enterprise software suite, helps with real-time inferencing of question answering, named-entity recognition and group classification tasks.
Accelerating Research Capabilities
By leveraging deep learning, life science organizations can accelerate their identification of key insights hidden in complex documents. For example, Vyasa’s Named Entity Recognition (NER) quickly indexes and categorizes key terms in data without the need for ontologies. This allows for the identification of key insights across grammar nuances, as well as semantic similarity. Vyasa’s NER will automatically discover new terms as they emerge in literature. For example, Vyasa began to identify COVID-19 before it was regularly appearing in organizational data.
This process is taken a step further with Vyasa’s Canonical Data Fabric, a collection of publicly available healthcare and life science resources. With the Canonical Data Fabric, users can connect to and easily search key sources such as ClinicalTrials.gov, PubMed, PubChem and others, which significantly expands their accessibility to key knowledge.
These capabilities accelerate key research tasks. For example, Vyasa recently worked with a partner to accelerate systematic literature review by 25%.
Enhanced Data: Smarter R&D
In a world where increased focus is spent on emerging diseases, access to data is key to enabling the innovation needed to improve life sciences R&D. With an intelligent data fabric, life science organizations can gain access to the data they need, accelerate data collection and analysis, and make more informed decisions that will lead to greater trial outcomes and improved therapeutics.
Vyasa will be showcasing its Layar intelligent data fabric for healthcare and life sciences at HPE Discover from June 28-30. Visit us at NVIDIA’s AI Pavilion in booth 202.
1. “R&D Pentathlon: Which Pharma’s R&D is faster, higher, stronger – ahead of the curve,” Cowen, 26 July 2021.