Cases of the Wuhan coronavirus have been confirmed around the globe, reminding us that the capacity to quickly develop, test and deploy new drugs is essential to save lives. Unfortunately, experts say we still are years away from a vaccine for the virus. However, a type of artificial intelligence can speed up the drug development process that already exists, but it hasn’t yet been deployed at scale.
How AI can improve pharmaceutical research
Most of us have heard of the more usual applications of AI in the healthcare sector, such as using deep learning for image analysis to improve cancer diagnosis. But there is a lot more AI can do, such as predicting the interactions between biological processes and drugs with extremely high precision. By leveraging the diversity of available molecular and clinical data, AI can identify molecules that have a high probability of being successfully developed into drugs that act on specific diseases safely and effectively.
AI can also help make clinical trials smaller, shorter, way less expensive but much more powerful. The technology can help identify trial patients much faster, reducing the total time from months to minutes. This is possible because it can quickly process large amounts of data, be they exam results or even doctors’ notes. AI is also faster and able to consider more factors when selecting patients for a trial than a human would be able to, from analyzing genetic information to targeting specific populations.
But things are not as easy as they may seem. Even if pharmaceutical companies know that big data represents a fantastic opportunity for the sector, they still struggle to make sure data is consistent, reliable and well linked – prerequisites for cutting edge analytics and artificial intelligence applications. That is because the industry relies on data systems that are outdated and difficult to use. As a result, data is often scattered across different departments and cut off from the rest of the organization. To increase the ability to share data, they must rationalize and connect these systems.
Semantic AI is a novel approach to building AI applications that can do just that. Semantic AI connects data across databases and links it in intelligent ways. By improving the quality of data, this approach to building AI ensures that machine learning algorithms have access to all the information they need to reach optimal results.
The benefits of Semantic AI: The AI that can accelerate the development of new drugs
One of the main applications of semantic AI is precise and agile data integration. It enables a comprehensive search for subsets of data on the basis of the links created, rather than on the basis of individual, distributed chunks of information.
Another advantage of semantic AI is that it stimulates and enables improved collaboration across teams. In the past, research and development in pharmaceutical companies were treated as activities that should take place behind closed doors, resulting in less internal and external collaboration. But at a time when pharmaceutical and healthcare companies are struggling to remain competitive and deliver solutions faster, isolation is no longer an option.
By connecting the different data sources that separate internal functions and improving collaboration with external partners, semantic AI can also help companies extend their respective knowledge and data networks across boundaries. And by ‘boundaries’ we mean different aspects – from language barriers to incompatible data formats or governance models that prevent collaboration.
Internally, the data integration brought about by semantic AI promotes collaboration between stakeholders in drug discovery, development, marketing and delivery. The result is knowledge and innovation across the entire portfolio. But the technology also enables integration with external data, allowing researchers to link internal research with external publications. For example, companies can benefit from the findings of the latest scientific breakthroughs in the academic field.
One of our customers, a global pharmaceutical company headquartered in Germany, already uses semantic AI to track and network the research activities of its global workforce. Initially, because of non-interoperable data sources, employees did not know exactly what other departments were doing, so they wasted time and effort duplicating research within the same area. They were also unable to accurately analyze their research focus in industry and needed new ways to link internal and external research and publication activities to control and monitor tasks. The company used semantic AI to transform a manually processed database into an automated tracking service with thousands of documents. Data from the R&D department is now automatically imported, providing real-time visualization and data analysis.
BenevolentAI, a company that aims to transform the way in which medicines are discovered, has combined machine learning and knowledge graphs to find better treatments for COVID-19. They have managed to identify a group of approved drugs that have both antiviral and anti-inflammatory properties. The drugs may prove to be essential in the treatment of the most severe cases of the disease, in which patients’ inflammatory response becomes a major cause of lung damage and subsequent mortality.
Getting started with Semantic AI
Companies interested in developing Semantic AI applications should first invest in building their Enterprise Knowledge Graph, which is a representation of human knowledge that can be understood by machines. A knowledge graph follows the human way of understanding the world by linking data containing all kinds of entities, business objects and concepts. It is the combination of the human ability to associate things with machines’ ability to be fast and precise which makes Semantic AI so unique and capable of delivering great results.
This type of initiative will eventually become a standard in the industry and in order to stay competitive, companies should be adopting it as soon as possible. The wealth of new data and improved analytical techniques they will acquire by combining knowledge graphs with machine learning will dramatically enhance their innovation capabilities.
