Joe Roberts, software developer here at IE, shares his thoughts on how breaking away from the traditional way of thinking can help your pharma company embrace artificial intelligence (AI).
The last few years have introduced everyone to AI. Techniques previously imagined only in science fiction are now in our homes and businesses. Industry leaders are jumping on the bandwagon, as experts demonstrate real world uses for AI showing the potential for vast growth and work flow changes.
On the brink of the 4th industrial revolution, do you understand the implications and applications of AI for your business?
Pop culture has influenced us to believe that AI will be the dawn of terminator style robots. This has caused real-life applications to be somewhat misaligned with what users expect.
The AI plateau is often implemented ‘under the hood’ and allows technology to adapt to the real world.
AI in the pharmaceutical industry
The pharma industry is a big driver of AI with companies like Sanofi already leading the way in exploring ways to speed up the discovery of alternative indications for existing drugs (drug re-purposing).
Johnson & Johnson are collaborating with IBM Watson Discovery, to read and understand scientific papers that detail clinical trial outcomes. They also use data scientists across Johnson & Johnson “…from strategy to process and governance as well as trading data as an asset and apply it in every single function across J&J from finance and HR” [ref].
Many start-ups are already exploring AI. BenevolentAI for example recently raised $115 million to grow their AI technology for motor neuron disease, Parkinson’s and other hard to treat disorders. They are applying AI across all phases from molecule discovery, to clinical trial design and analysis.
Why use AI in business?
“There is real potential for AI throughout the industry. It is not a remedy to all business model problems, but rather a tool for deeper insight into huge lakes of data for guidance.”
Incorporating an AI layer into your business won’t necessarily provide instant return on investment, or improved department by department productivity. What it can provide is an invaluable analysis on model state, hidden bottlenecks in outdated models such as Chain-link systems, and open opportunities to automate business procedures to potentially gain market leverage.
Most pharmaceutical companies store millions of compounds to screen for potential drugs. This screening process is slow and yields little results, even with the use of robotics and lab automation tools. Using past data sets of drugs and trials, machine learning tools can build virtualised libraries using the chemical structures and properties from the datasets for new compounds that could dramatically accelerate identification of drug leads.
What is AI?
The term artificial intelligence is actually a very broad one as AI is made up of a subset of concepts. There are 3 main basic concepts to understand here – machine learning, deep learning and neural networks.
In the simplest of terms, a machine learning algorithm is based on mathematics and can vary from a single linear regression to a multi layered algorithm which we call a neural network.
Deep learning is a more advanced implementation of this using very large, deep layered neural networks. These neural networks consume huge amounts of data (or big data) to create forecasts or classifications in areas such as image recognition, patterns in drug trial data, or market buying trends.
Applying the infrastructure
“The key to AI is data. Start storing the right data, and plenty of it.”
Most businesses sit upon goldmines of data and are unaware of the value it holds. In preparation, it is useful to note that not all data is good data, and in such an early stage we wouldn’t necessarily know what data we would need for later use when building a machine learning model.
If there are existing custom in-house systems and database management systems in place, a good practice is to prepare the infrastructure. Clean out discrepancies and ill structured relationships that cause redundant data within your databases and/or applications. This is in fact part of a process of data science in which we prepare the data to pass through a machine learning model.
But most importantly, start asking yourself “What do I want AI to achieve?”.
Here at Innovative Edge, our development team are keeping up with the latest trends in AI to understand and discuss ideas about how we can strategically incorporate exciting new innovations and technologies into our Ignite and LEx Catalyst technology. So watch this space!…
To find out how our technology solutions can help you optimise your launch planning, please contact us today for a chat or to organise a demo on +44 (0)1892 825110, firstname.lastname@example.org, or visit www.innovative-edge.com.
Joe Roberts is a Software Developer for our Ignite Launch Tracker Platform. Joe works to bring new features, as well as maintaining and continuously developing the platform. With a keen interest in AI, he is very enthusiastic about applying new technologies.