A guide to using artificial intelligence in Pharma

What is Artificial Intelligence?

“The theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making and translation between languages” English Oxford Living Dictionary

Artificial intelligence (AI) is a field of computer sciences that seeks to create intelligent machines, to replace or augment the human intelligence ability.

This form of technology excels at the ability to assess a significant scale of information and spot patterns and trends that would be beyond the ability of a human researcher. In this process, knowledge engineering is key. Artificial intelligence machines can only reach decisions and share logic if they are programmed in the correct way and fed enough relevant data.

Currently, many AI systems are focused on problem solving and pattern recognition, with a majority looking at a very specific goal. There is also a wide range of applications of AI, ranging from those that simply augment a human’s ability during a decision making process to machines that are more autonomous in nature.

“[We’re seeking to] create smarter, more useful technology and help as many people as possible” Google AI

How is AI being used in the life sciences industry?

In recent years, the life sciences industry has started to invest heavily in AI, partnering with a number of specialist companies. One of the main drivers behind this AI adoption is a hope that by using the technology the industry can address the declining drug productivity rates.

The life sciences industry is facing pressure on two fronts. Firstly, the resources needed to develop a new drug are on the rise, yet the success rates and profitability are falling. At the same time, there is more demand for targeted and personalized medicine which requires a much more technical development approach. To meet these challenges, the industry has had to explore new technology.

With many companies storing a substantial amount of research data across many decades, AI may be able to find new patterns and connections in this treasure trove and uncover new potential candidates for drug development. The scope of data is currently beyond human comprehensive, but with the right programming, a machine may be able to bring about the next breakthrough in science.

Some current examples of how and why the industry is using AI include:

• Astellas is working with Biovista to focus their AI efforts on repurposing existing compounds.
• AstraZeneca and Berg Health announced a partnership to discover therapeutic targets for neurological diseases. They’ve also partnered with BenevolentAI to focus on chronic kidney disease and idiopathic pulmonary fibrosis.
• Bayer has used Cyclica’s technology for off-target effect investigation and multi-targeted drug design. They’ve also partnered with Sensyne Health to collaborate on the development of an AI solution to find treatments for cardiovascular disease.
• GSK has one of the largest teams dedicated to AI. They have partnered with Exscientia and Insilico Medicine to discover novel and selective small molecules. Earlier this year, they announced that the partnership with Exscientia had produced the first tangible result, which was a highly potent lead molecule targeting a novel pathway for chronic obstructive pulmonary disease.
• Janssen partnered with BenevolentAI to identify untapped potential in their portfolio. They’ve also explored projects to predict neurodegenerative diseases from voice samples and use virtual drug design technology.
• Merck have also partners with Cyclica to be able to use their AI-augmented proteome screening platform. And they have partnered with Iktos to use a generative AI system in the design of novel molecules.
• Novartis has built in-house AI capabilities and partnered with start-ups, large companies and leading academics. They are currently pursuing research into predicting how patients will respond to drugs and improving the operational efficiency of their clinical trials.
• Takeda have partnered with Recursion to evaluate and identify novel pre-clinical candidates for rare diseases. This has led to new therapeutic candidates for more than 6 diseases.

What do the Pharma IQ experts think about the use of artificial intelligence in pharma?

We have spoken to our industry experts to gather their thoughts on the future of artificial intelligence in pharma, including where the industry will face limitations and how to best use this nascent technology.

Don van Dyke, Chief Operating Officer, Cloud Pharmaceuticals

Earlier this year, during our Smart Labs Digital Online Event , Don van Dyke shared the Cloud Pharmaceuticals strategy to find novel compounds through the use of artificial intelligence.

He believes that “using a targeted drug design logic and process allows for a greater probability that the molecules designed will progress through the drug development process”. This is distinct from the uncertainty faced with traditional methods, where productivity and profitability continue to drop.  

By adopting an AI approach, Don expects that companies will be able to prosecute more targets in a shorter period of time. He said this will “lead to a change in the sort of metrics that the pharma companies, start-ups and biotechs use, when they now have an agile, fast and inexpensive way to go after their targets”. If companies are able to go through targets in a quicker fashion, with less investment needed, they will be able to be less risk averse.

In a recent analysis for Pharma IQ, Keinan, Shipman and Addison of Cloud Pharmaceuticals shared that while computational drug design is achieving industry recognition, it is still in the active research phase. They argue that to be successful with the use of AI algorithms, expertize is needed in chemistry and biology to curate and analyze data-sets. It’s also important to understand the questions being asked through this process and the relevance of these questions to the available data. They also highlighted that a critical task is to avoid discarding or ignoring information from prior analysis and modelling.

The team believe that at the present time, “the best application of AI in drug discovered is ‘augmented intelligence’… the combination of the best from both AI and computational chemistry tools especially shines in cases of limited or inconsistent data”. It is in this niche where AI could be a revolutionary tool.

“Some people call this artificial intelligence, but the reality is this technology will enhance us. So instead of artificial intelligence, I think we'll augment our intelligence.” —Ginni Rometty

Praful Krishna, CEO of Coseer

Praful is keen to point out that the landscape in life sciences is changing. To navigate these new waters, pharma giants will need to embrace new tools, including artificial intelligence. 

Many are convinced that the era of blockbuster drugs is over, with focus now on personalized and precision medicine. But to adapt to this change pharma companies will have to re-tool themselves and move away from their traditional approach.

As Praful notes “despite all of their resources and expertize, pharma giants won’t be able to adjust their business models overnight. If plucky biotech start-ups ripe for scale-up can boost declining pipelines, then acquiring tech innovations like AI could be critical”.

Praful believes that AI will be most useful when it can move beyond gesturing in the direction of a possible answer and provide practical insight that can actually improve the life of a scientist.

“The playing field is poised to become a lot more competitive, and businesses that don’t deploy AI and data to help them innovate in everything they do will be at a disadvantage.” — Paul Daugherty, Chief Technology and Innovation Officer, Accenture

Siniša Belina, Senior Life Sciences Consultant at AMPLEXOR Life Sciences

Siniša argues that AI and machine learning are a natural part of the broader transformation of the health value chain. To him, these technologies truly have the potential to “accelerate scientific breakthrough, identify previously elusive patterns in unwieldy global data masses and enable greater drug personalization”.

To achieve this he does believe that it will be crucial for humans to oversee the process and sense check the results. His ideal scenario is to let IT systems take over the operational loads and allow the experts to do the most interesting and mentally demanding tasks.

He also sees potential for artificial intelligence beyond the drug discovery process. Although this has been the focus of many industry efforts, he believes that by using AI we could expect “new advances in medical imaging interpretation, genomic profiling, personalized medicine and new treatments”. By combining this with the healthcare value chain and new layers of ongoing patient monitoring, the industry will be able to develop a keener strategy for more pre-emptive, preventative care. This will allow for a “shift towards maintaining wellbeing rather than reacting to illness” which could radicalize the way we approach drug development and treatment plans.  

Clay Heskett, Ben Faircloth and Stephen Roper of L.E.K. Consulting

L.E.K. Consulting shared with us that in a recent survey, they found that the majority of respondents believe that AI applications will become standard in the pharma operating model over the next five to ten years.

Although, they recognize that presently “the landscape of AI providers and technologies is fragmented, with no clear winners in any application”.

For those looking to move forward with an AI implementation, L.E.K. offers three important considerations.

1. Consider a partnership with an AI company – Considering the scarcity of talent who have crossover skills between AI and biology, it will be expensive and challenging to build in-house expertise. L.E.K. believe that “a more effective strategy for many companies will be to build partnerships with leading AI companies” and leverage their experience
2. Be transparent with regulators when it comes to your algorithms – As L.E.K. explains, “without transparency, AI risks becoming a ‘black box’”. Without rigorous testing and review, there may be unforeseen issues when drugs go for regulatory approval. Bringing regulators into this process early on can limit these issues
3. Be mindful of data privacy – L.E.K. make clear that “the use of patient data is highly sensitive, so, as AI capabilities develop, companies must take the appropriate legal and compliance measures to protect the increasing volume of such data”. Considering the growing scrutiny around data protection, failing to safeguard data could be a serious issue