In the first of the Pharma Integrates Insights series, Marco Mohwinckel examines the adoption of artificial intelligence (AI) by pharmaceutical companies and sheds some light on what’s real … and what’s yet to come
AI is big business! The mere mention of artificial intelligence helps you to raise money, get promoted and generate a raft of social media likes, retweets and followers. It’s popular and, by default, makes you popular too.
The fascinating paradox behind AI is that it has suddenly burgeoned to become both the disruptor and saviour of all industries. Indeed, it seems that every sector, no matter what the technology level, is touting AI as its next paradigm shift. The hype around AI is understandable; its potential is truly incredible and beyond most people’s current comprehension. However, contrary to most reports, AI is neither new nor a panacea for all ills.
AI first appeared at a computer conference in 1956, when John McCarthy, Professor of Computer Science at Stanford University, postulated that we should “proceed on the basis of the conjecture that every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it.”
Sounds impressive … but why then did it take more than half a century for AI to truly explode? The answer can be found at the intersection of a dramatic increase in computing power and the widespread adoption of technologies that enable both an ever-growing interconnectivity and exponential data generation. Experts disagree as to whether AI will replace human intelligence; and, although AI is certainly “artificial,” should we not ponder just how “intelligent” it really is? Should we refer instead to “smart or sophisticated algorithms underpinning predictive models”?
Big Data: The Lifeblood of AI
Without meandering into philosophical discourse, the principle of “garbage in, garbage out” still resonates. AI or, indeed, any algorithm, needs to be built, then trained and calibrated using data. The more data, the better. So, in our algorithmic and AI-driven world, data is the new oil. Thus, anyone who generates, owns or controls large and growing datasets has a real, sustainable and self-reinforcing competitive advantage.
Indeed, the more we use technology, the more data is generated, the more insights we can derive, the more predictions we can make and the more specific behaviours we can activate. This, in turn, drives technology use, generating even more data … and so on. No wonder then, that in our knowledge-turned-platform economy, tech giants, whose products and services billions of people use every day, are the most valuable companies in the world.
Interestingly, unlike oil, data is not a scarce resource. It’s widely understood that scarcity drives value; so, when supply is bountiful — the more data the better (for AI at least) — quality counts. Quality is crucially important, especially when we move from the trivial (recommending films, groceries or make-up) to the profound (informing patients and doctors about decisions on health and disease).
A vast amount of public and proprietary healthcare data is available, from scientific literature, publications, clinical trials, patents to guidelines, presentations, social media, EMR/EHR, genomics and bioinformatics, etc. However, much of it is unstructured, fragmented, siloed and, often, taken out of context. Experts agree that much effort is still required to cleanse and reconcile scattered and patchy data, which is still a very labour-intensive endeavour.
In fact, many healthcare data sets, assuming they are accessible, fail to meet minimal market research requirements, let alone regulatory standards. It turns out that in healthcare, our AI-friend still needs a lot of help from medical experts to articulate ontologies, define research questions and master everything from programming, testing and fine-tuning algorithms to deriving insights. After all, it’s going to take some time to breath 4 million years of human evolution into a machine.
Far from “escaping the lab” and having a life of its own, healthcare AI currently depends on human intelligence to augment it. We’ll no doubt see and welcome an increasing number of powerful predictive healthcare tools that can empower both patient and physician, as long as they can use high quality data with discretion and integrity. Concerned with this matter, the UK government has recently published the “Code of conduct for data-driven health and care technology,” defining the responsible use of healthcare data. It’s a good framework that private enterprise should engage with and improve.
Is Conservative Pharma Embracing AI?
The pharmaceutical industry is often accused of not being innovative, of being conservative and a slow adopter. Yet, the contributions that drug companies have made during the past century to advance the science of medicine and human health could not have been done without a highly innovative mindset and significant bets on cutting-edge technologies. Even today, pharmaceutical organisations continue to invest significant amounts of money, under highly uncertain circumstances, into finding new treatments and cures. Drug discovery is messy and drug development is expensive, very expensive. If our ambition to treat and cure diseases, promote health and wellbeing has no limit, there is no doubt that research-based pharmaceutical companies will continue to play their fundamental part in fulfilling it. Society needs Big Pharma.
If indeed pharmaceutical companies are more innovative than most people give them credit for, what is their stance on the most hyped technology of them all? Is pharma embracing AI? Where, how, why and to what extent? The short answer is that pharmaceutical companies are starting to use AI across the value chain, with a substantial emphasis on drug discovery. In fact, a 2017 survey by the Pistoia Alliance, a global, not-for-profit alliance that works to lower barriers to innovation in life sciences R&D, found that most pharmaceutical companies are either using or integrating AI into their operations.
It’s arguable that some pharmaceutical companies have become “too big to innovate,” but all seem to recognise that the decline in R&D productivity can be reversed by embracing open innovation models, new collaborations and partnerships, and by opening the door to entrepreneurs. In drug discovery, pharma is embracing AI to “unlock” human biology, to generate new research hypotheses and discover new targets. Drug discovery is, more often than we care to admit, a serendipitous process. Researchers are hoping that, by combining AI’s power to sift through vast amounts of scientific data with their deep ability to make sense of it, they can reduce serendipity and deliver the cost-efficient breakthroughs that the industry badly needs.
But the use of AI in pharma is not limited to drug-discovery. It’s also applied to test new drugs in silico — through simulation and modelling — to improve the efficiency of development processes and supply chains, and to substantiate the value of new medicines in the real world.
And although pharmaceutical companies remain keen on embracing AI, they will proceed with characteristic caution. They need the proof that AI can actually accelerate discovery times, save money or reduce complexity before investing and/or building significant in-house capabilities. Indeed, the question remains regarding how much should be built versus bought, licensed or accessed through partnerships. More agile start-ups may be better placed to push AI-driven drug discovery forward; but, if pharma chooses to build capabilities in-house, how attractive will their conservative and hierarchical environment, with all the checks and balances, be to those disruptive post-millennial data scientists it badly needs?
In this interview series, we will answer some of these questions and dig deeper into the use of AI in drug discovery. We will also touch on other disruptive technologies, from blockchain to virtual reality, by juxtaposing the dreams and aspirations of entrepreneurs to the more grounded perspectives of pharmaceutical executives and, with luck, build a tangible bridge between hype and reality.
Pharma | Digital Health | Health Tech | Strategy & Innovation