How Big Pharma Adopts AI To Boost Drug Discovery

(Last updated: July 2023)

The type of artificial intelligence (AI) which scares business leaders, experts, and activists all over the world, is called “general artificial intelligence"—the one which could “think” pretty much like humans do, and which could quickly evolve into a dangerous “superintelligence”. There is a notion that it might be invented in the nearest decades, but today we are definitely not there yet. However, with the recent groundbreaking advances in deep learning and natural language processing technologies, particularly -- large language models (LLMs), we have all felt that the world might indeed be changing more rapidly than AI deniers used to think. Let’s face it, only few could foresee such an alarmingly efficient public release of the most generalized AI model of all time -- ChatGPT, by OpenAI. Adding more to that, a race of LLMs has begun, with Google launching Bard, and other companies following the path.

Even notoriously slow for technology adoption, the pharmaceutical industry has seen accelerated integration of various AI technologies over the last decade, and the interest is rapidly growing. The potential impacts of this transformation extend beyond healthcare providers and patients grappling with difficult-to-treat ailments, reaching into the biotech sector as well. Based on projections from Morgan Stanley Research, even slight enhancements in early-stage drug development success rates, facilitated by artificial intelligence and machine learning, might result in an additional 50 innovative treatments over the next decade. This could equate to a market opportunity exceeding $50 billion.

According to a 2022 report by GlobalData, 50% of professionals within the healthcare industry would prioritize investments in AI over other emerging technologies, such as big data (38%), digital media (37%), cloud computing (31%), real-world evidence, RWE (27%), and others.

The 2022 thematic research report titled 'Artificial Intelligence (AI) in Drug Discovery' from GlobalData predicts that the total expenditure on AI by the pharmaceutical sector is projected to escalate to more than $3 billion by 2025.

Let’s review specific examples of how AI is used in the pharmaceutical industry.

(Since most AI-driven companies use a mix of different approaches and rely on interdisciplinary sources of data for their modeling work, the below classification of AI use cases is illustrative.)

AI for drug target discovery and disease modeling

One of the most promising areas of AI in pharma is modeling biological systems, and identifying novel drug targets. A number of AI companies, such as CytoReason, are specifically focused on building advanced disease models, for example.

In March 2023, AstraZeneca presented preclinical data on an AI-generated target, the Serum Response Factor (SRF), for idiopathic pulmonary fibrosis (IPF) -- from its collaboration with UK-based AI company BenevolentAI. The target, discovered via BenevolentAI's AI-enabled drug discovery engine, underwent thorough experimental validation by AstraZeneca, involving CRISPR screening in primary human lung fibroblasts and validation via SRF gene silencing or pharmacological SRF pathway inhibition. The presented data indicates that inhibiting SRF-driven transcription of pro-fibrotic genes in lung fibroblasts could potentially lead to antifibrotic efficacy in IPF. To date, the collaboration between BenevolentAI and AstraZeneca has resulted in five AI-generated targets selected for portfolio entry, three of which are for IPF. This successful partnership was expanded in January 2022 for another three years, including two new disease areas - systemic lupus erythematosus and heart failure.

Several months earlier, AstraZeneca announced a strategic research collaboration with Illumina, a global pioneer in DNA sequencing and array-based technologies. This collaboration aims to expedite drug target discovery by melding their respective competencies in AI-based genome interpretation and genomic analysis. The initiative will examine if a unified approach utilizing these technologies can bolster the efficiency and certainty of target discovery in pursuit of promising drugs built upon human omics insights. AstraZeneca's Centre for Genomics Research will adopt a framework merging the AI-based tools of both companies, leveraging next-generation AI interpretation tools like Illumina's PrimateAI and SpliceAI, along with AstraZeneca's own tools such as JARVIS and in silico predictors.

In September 2022, Pfizer the expansion of its multi-year partnership with Israel-based AI in pharma company CytoReason. Under this agreement, Pfizer will invest $20M in equity, with the option to license CytoReason's platform and disease models and fund further project support in a deal that could reach up to $110M over the next five years. Since the initiation of the collaboration in 2019, Pfizer has utilized CytoReason's biological models in its research to boost the understanding of the immune system for the development of drugs for immune-mediated and immuno-oncology diseases. This additional investment will aid the development of more disease models and the creation of high-resolution models across various therapeutic areas.