The future of drug discovery is AI. However, pharmaceutical companies risk being left behind if they are unable to make a persuasive case for AI investment
18 February 2019
In December 2018, Alphabet’s artificial intelligence (AI) firm DeepMind successfully predicted a series of 3D protein structures at the 13th CASP contest, a global competition to tackle the age old puzzle of protein folding prediction. DeepMind’s ‘deep learning’ paved the way for a solution to effectively diagnose and treat protein misfolding diseases such as Alzheimer’s.
A growing number of AI driven companies are using similar methods to tackle complexities in the drug discovery value chain. The pharmaceutical industry’s investment in AI is also rapidly gaining momentum. The last two years have been marked by new collaborations between pharmaceutical companies and AI driven startups. The gaps between tech startups and established pharma companies are increasingly becoming narrower with the advent of interdisciplinary AI and related technology offerings.
AI’s role in drug discovery is evident in three major use cases:
- Target and Biomarker Identification: Combining AI and machine learning with plethora of data sets opened up new opportunities in target and biomarker identification. For example, Roche is using GNS healthcare’s casual machine learning and simulation AI platform to generate insights for cancer drug development from longitudinal electronic medical records, next generation sequencing and omics data. Similarly, Sanofi Pasteur and BERG partnered together to identify influenza biomarkers using BERG’s AI platform and Sanofi’s trials data.
- New therapies: AI is being used to design and deliver high productivity drug candidates. GlaxoSmithKline (GSK) has entered into strategic collaboration with Exscientia to discover small molecules using GSK’s discovery expertise and Exscientia’s AI platform. Similarly, BenevolentAI is using its AI platform to bring new therapies for Amyotrophic Lateral Sclerosis (ALS)
- Drug repositioning: AI is helping pharmaceutical companies to derive additional benefits from existing assets, so the early drug development can be bypassed with potential cost savings. BenevolentAI is using Johnson & Johnson’s products to reposition in new indications using its AI driven platform.
Although AI is generating a huge buzz, pharma executives need to carefully consider its business implications and it is also hard to justify the return on investment. However, companies might get left behind in this race if they are unable to make a persuasive case for AI investment.
Things to consider when making a case for AI investment:
- Opportunity vs Need: AI involves broad range of tools and technologies, from machine learning to probabilistic reasoning. Companies should go for the right technology portfolio depending on their scientific needs and therapy area of focus. For instance, a company trying to aggregate and mine information to understand the underlying correlations in fragmented datasets with incomplete information may invest in a deep learning AI algorithm. Similarly, a company looking to identify hypothetical drug targets by analysing RNA sequencing data may want to invest in machine learning, cloud computing and big data techniques.
- Doing it alone vs Partnering: Building in-house AI capabilities requires four components: data, algorithms, computing power and people. In most cases, pharma has the required data and AI companies have the required technologies and resources. Pharmaceutical companies can work with these AI companies to outsource or collaborate on discovery projects. However, there might be instances where it makes more sense to go that extra mile to build in-house AI capability. If a company owns exclusive datasets that have clear competitive advantage, then in-house AI investment is a strategic business need. The new AI architecture should be tailored to the specific prediction problems or to the target data sets.
- Data before AI: Big data is a prerequisite for AI and the technologies continue to mature due to the explosion of data in the recent years. Without quality data, AI would be of no use. Data is more important than which AI technology you use and the industry should invest in generating quality datasets to feed into the AI architecture.
- Heading up AI initiatives: AI is not always about computer technology and IT expertise. Either building in-house or collaborating with an external AI vendor, the team leading these initiatives should comprise business, technology and scientific leaders.
Many AI technologies and tech companies have emerged in the recent years and the pharmaceutical and life sciences industry can no longer ignore the impact of Industry 4.0. AI offers pharma an opportunity to accelerate the drug discovery and to explore new ways to diagnose and cure diseases.
At PwC, we believe in helping our clients develop successful long-term AI strategies that are in tune with a responsible approach that benefits society. Find out more: pwc.co.uk/ai