AI & Machine Learning Paradigms: How life science companies are implementing AI-driven analytics to enhance their operations and drug development

While artificial intelligence (AI) has been around for decades; it has only recently become the buzzword of choice in the tech world, with conjecture surrounding its capabilities running rampant. AI is concerned with building machines and/or processes that would typically require human intelligence. AI is an umbrella term which encases not just one technology, but a collection of them which will later be discussed in further detail. AI is rapidly becoming extremely prevalent in the pharmaceutical industry; we are here to discuss trends and potential benefits that AI offers namely in the research & development/drug discovery life cycle.

For such an important industry, roughly 90% of what the pharmaceutical industry seeks to accomplish in clinical development ultimately fails (Hingorani, 2019). AI and its related technologies and practices have the potential to uniquely transform many aspects of the pharmaceutical sector to essentially create higher success rates. The research and development processes are of utmost importance in the pharmaceutical industry and the costs reflect just that. The costs associated with the development of a new drug are staggering, often surpassing the billion-dollar mark in addition to a 12-year average timeline. Profits on massive R&D investments for pharmaceutical companies are contingent on FDA approval, which is very difficult to achieve. According to a study from MIT, a mere 14% of drugs in clinical trials win FDA approval (Hale, 2018). Pharmaceutical companies are beginning to implement the use AI and machine learning to phase out unsuccessful drugs/therapies prior to incurring overwhelming costs.

The COVID-19 pandemic has especially highlighted the need to create and develop effective drugs and treatments. COVID-19 vaccine and treatment pioneers like Moderna, Pfizer, and Regeneron have led the pack in their use of AI and its related automation practices. At a high level, automation can improve laboratory efficiency, and lower overall attrition, all the while decreasing costs and shortening development and rollout timelines in the process.

Now this begs the question – what kinds of AI related efforts are these pharma giants engaging in? The key for pharmaceutical companies is to reduce the time it takes for the drug to reach the market. The artificial neural network is the most prominent AI application that is used in drug design and discovery; they are the basis for what is referred to as deep learning, a subfield of machine learning. To understand the artificial neural network, it is important to first understand what machine learning seeks to achieve at a high level – to use and learn algorithms without reprogramming. Neural networks are modeled after the neural and synaptic structure of the brain; the neurons are arranged in layers which are used to learn the input data and make a prediction. Data is currently the hottest commodity on the market; to gain a competitive edge companies must make use of this data to the best of their abilities. Neural networks have proven to be one of the most sound and reliable unsupervised learning tools for large data sets. While pharmaceutical companies are presented with massive datasets, they need to employ the proper AI tools to appropriately analyze and extract valid statistical models.

Life science organizations ranging from clinical to advanced commercial stage companies can implement AI-driven internal IT structures to enhance their business operations and future drug development. Here at Stratos, we partner with life science companies to perform master data assessments on their current data structures to ensure their current environments are optimized for maximum efficiency and security, and scalable for the future as they grow. This ensures their systems are designed to load the massive datasets required for AI driven reporting and analytics.

AI can be utilized to mine and analyze any potentially relevant biomedical papers and articles to for instance, identify the best molecules and compounds that will provide the highest success rates in a certain clinical trial or drug discovery undertaking.      

AI-based algorithms are becoming more and more technically advanced in their capabilities by the day. The availability of large datasets and peer-reviewed medical papers are driving the growth of AI productivity. Machine learning and other AI algorithms have shown tremendous promise in the drug discovery space and their utility has been demonstrated convincingly. The limitations of AI are seemingly endless, in my opinion we are still in the first inning of this game.

Article by Iordanis Kapanides