by CP Dargar, Life Sciences Program Director at TATA Consultancy Services
This blog originally appeared on the TCS Life Sciences Pulse blog pages.
Breakthrough drug approvals reached a 21-year high in 2017 after the US Food and Drug Administration (FDA) cleared 47 new molecular entities (NMEs) for the market. While pharmaceutical R&D has never seen better days, numerous toll gates in the form of manual decision making stages across the product development lifecycle continue to create bottlenecks.
Labs, trials, and the compliance paperwork together push the cost of discovering an NME up to USD 2 billion or more. By the end of a decade when the drug is ready to hit the shelves, enterprises are forced to price it so high that it becomes unaffordable for most consumers, especially in low income countries.
The volume of data being generated by genomics research per day is doubling every seven months – comprising critical dosage and formulation information along with trial safety signals. The current IT infrastructure is incapable of handling this flow, let alone store and share it securely. Compliance requirements, including the ones recently introduced by the General Data Protection Regulation (GDPR), make this even more difficult.
Unlocking the Human Genome
Having recognized this problem, organizations such as St. Jude Children’s Research Hospital and The Ontario Institute of Cancer Research (OICR) are developing cloud-based platforms as clinical and medical data repositories with embedded analytics capabilities. These solutions are expected to help researchers set up a single source of truth (SSOT) for informing the entire drug development process. Such platforms will eventually become the foundation for artificial intelligence (AI)-led process automation and machine learning (ML) programs for supporting clinical decisions and predicting the effectiveness of a new drug even before it reaches the trial stage.
Through context normalization, AI dramatically increases the quantity of data that can be analyzed in the course of the drug discovery and testing process. In turn, it can simultaneously generate and unbiasedly test new hypotheses at a rate that would be impossible for human researchers. By visualizing the resultant insights through a centralized dashboard, companies can not only expedite the decision making process at every stage, but also gain complete visibility across the R&D value chain.
Pfizer’s AI platform, for example, can analyze private data such as lab reports and helps researchers identify potential relationships between drug action and diseases using dynamic visualizations. Once an AI-based solution such as this is fed with enough historic data from previous and concurrent trials, it can further minimize the need for conducting animal studies through in vivo simulations that help pinpoint molecules for taking forward to the live testing phase. Furthermore, it will be able to discover safety signals that may have otherwise been lost in the steady stream of information generated during Phase II trials.
Accelerating Drug Discovery
AI’s role doesn’t end there. In combination with natural language processing (NLP) and deep learning algorithms, it can increase R&D success rates significantly by generating and validating a pipeline of clinical hypotheses. A platform developed by BenevolentBio can generate as many as 36 and validate 24 out of them in vitro compared to a typical lab researcher who can manage to produce about five of them in the same time period.
This is just the tip of the iceberg in terms of what AI can do in the life sciences space. With consumers coming to openly embrace wearables and mHealth apps, companies are now in a position to capture clinical data from patients as they go about their normal lives. It can then use predictive criteria along with this real-time data feed to determine the outcome of taking a drug, whether a patient will drop out, and if a trial will be successful. Clinicians can review a patient’s medical history and possibly even receive lab results within one system simultaneously. Are we then nearing a future where pharma R&D will be a collaborative effort between human and machine?