How Artificial Intelligence will Transform, Rethink & Augment the Drug Development Process in Life Sciences

22 January 2020, 12:53

In this blog, we will highlight and explain how artificial intelligence will transform, rethink and augment the drug development process in the Life Sciences industry. 

As described by McKinsey & Company in November 2019 (article), the estimated average cost of bringing a drug to market (including drug failures) is now approximately $ 2.6 billion, a 140% increase in 10 years. Given the continuous release of new state-of-the-art algorithms (by e.g. universities, expertise centers, etc.), the huge amounts of (un)structured clinical data as well as the nature of the Life Sciences industry (expertise/processes), the role of artificial intelligence will only increase as from 2020. Artificial intelligence will not only be used to augment the drug development process in Life Sciences but also to challenge, transform, and rethink the current way-of-working. 

Innovations in biomedical sciences and technology fuel the opportunity to transform R&D for new-drug development holistically — 500 days faster, better tailored to patient needs, and 25 percent cheaper.

– McKinsey & Company (November 2019; article)

This blogpost will highlight and explain the significant role of artificial intelligence in optimizing the new-drug development process by doing it 500 days faster and 25 percent cheaper (cf. McKinsey & Company quote). 

DISCLAIMER: Please note that implementing and deploying artificial intelligence within the drug development process will never be a one-off exercise, but rather a continuous process between the self-learning algorithms (artificial intelligence) and clinical subject matter experts (clinical researchers). 

Designing and developing a clinical trial protocol

During the research, design and development of a clinical trial protocol, researchers need to gather, use and digest a vast amount of data. On the one hand, this leads to the fact that this is an extremely expensive and time-consuming process while on the other hand, researchers cannot see the forest for the trees. One of the most significant characteristics of artificial intelligence is its ability to digest this vast amount of data and identify new patterns and insights. Below, you can find a subset of documents (clinical research papers, doctor reports & patient documents) that are being used and analysed in order to design the clinical trial protocol for a new drug to test. 

One of the most promising domains of artificial intelligence is Natural Language Processing (NLP) and Natural Language Understanding (NLU).

Within these domains, self-learning techniques can be applied in order to on the one hand understand the rationale in each of these documents, while on the other hand also being able to extract certain fields/entities (Named Entity Recognition). Given the fact that these self-learning algorithms could be applied on thousands (even millions) of documents, a multi-dimensional clinical graph could be extracted and developed. The core of this clinical graph will be based upon three main concepts: Patient (demographics & medical history), Disease (medical diagnosis) and Medication (medication composition, – usage & – impact). 

This multi-dimensional clinical graph could be used to e.g. analyse on which (medication) compositions/dosage to research for which diseases and e.g. which patient population could be used to test the new drug. This graph enables the clinical researchers to transform, rethink and augment their clinical trial protocol design process (cf. output: clinical protocol document below). 

In addition to designing the clinical trial protocol and identifying which patient population to target your new drug to test (cf. section above), self-learning algorithms can also be used to monitor the effect of the drug on the target population during the clinical trial itself (cf. monitoring – and evaluation process). These types of engagements typically have two phases: data gathering and analysis phase. During the data gathering phase, typically the following data sources are being used: doctor notes and laboratory reports (NLP/NLU), health metrics captured by wearables and input from HCPs (time series), etc. Consequently, the self-learning algorithms will be able to identify new patterns and insights in these data and spot (potential) anomalies.

In conclusion, artificial intelligence is – and will play – an (even more) significant role within the drug development process in the Life Sciences industry as from 2020. Feel free to contact us if you are interested to see and hear more on this. 

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