Is Artificial Intelligence Able to Find the Next Holy Grail Within the Drug Research & Discovery Segment of Life Sciences?

28 January 2020, 12:21

In this blog post, we will clarify how Artificial Intelligence can benefit the life science industry within the scope of drug research and discovery.

Numbers have shown that only 10% of the drug candidates enter the market. Combine this with the time consuming (up to 15 years) and costly process and it is clear that every improvement along this journey results in a big impact for multiple stakeholders.

This blog post will focus on the first stages of this process, up to the pre-clinical trials. In a follow up blog post, we will cover the subsequent phases of clinical trials and the related manufacturing & sales/marketing aspects.

Drug discovery & research gets kick-started by finding a potential drug candidate, which is already a tough nut to crack. Researchers put a lot of effort in literature study of papers and previous trials, to find that one specific drug candidate. After crawling through a lot of structured and unstructured information, available in numerous open source databases (e.g. UniProt ), they end up with a drug candidate or no candidate at all. 

This is where Machine learning kicks in: With Machine learning we can combine all these sources to support a researcher, by clustering the textual and visual related content and provide him/her with the most relevant content. For ease of understanding, this can be seen as a recommendation system for (bio-)medical information.

Nevertheless, the process of finding a potential drug potential candidate with AI is still at the “learning to crawl”-stage. Therefore, the challenge will be twofold in order to actually get to valuable insights, plus convincing the scientists/research of the given prediction.

These insights could also contribute to chemical biology, boost the design of more effective molecules rather than rely on empiricism to discover them.

Once we select a potential drug candidate, we can start looking for e.g. a molecule which has the right properties to bind with the identified target. Again, Machine Learning can operate as a matching engine between the target and the molecule, to predict binding in the subsequent stages of the process.

From there on, we tend to start testing the effectiveness on animals and in a later phase, on human beings. These (pre-)clinical trials are very costly in terms of time and money. Therefore, the higher the success ratio of these stages, the higher the ROI. By using cautiously trained machine learning models, we can predict effectiveness, toxicity, and other numerous potential effects without having to run any tests on animals or human beings.

To summarize, AI can support the drug- and research process in numerous ways and can boost the success ratio and related ROI.

More information on how to apply AI in the life sciences industry can be found on