An explosion in scientific data has transformed the way researchers, across the sciences, access insights to form and test their hypotheses. It’s become almost impossible for an individual researcher, or even a team of researchers, to keep up with the constant influx of information.
This is no different in pharma, yet the process for scientific discovery in the field has hardly changed over the past fifty years. Rapidly advancing computing power and developments in artificial intelligence (AI) present us with a real opportunity, however, to transform the way we identify and validate our research. Using this technology, we can readily access insights to inform hypotheses, and develop more efficient processes and better treatments for patients.
Unlocking the insights
The pharma industry often falls under scrutiny for its inefficient practices. Despite the deployment of advanced technologies such as high-throughput screening and biomarkers, the cost of bringing a drug to market remains high and the success rate low. In fact, the overall success rate to develop a new medicine remains in the 5% range, while costs have soared to around $3bn per product launched.
To ensure the most suitable drug candidates are being put forward for clinical trials, researchers need to access more data and insights. However, this remains difficult when the average academic researcher only gets through an estimated 250-270 articles per year. To put this into context, there are over 50 million scientific articles worldwide. This means there is an enormous gap between the data available and what researchers can take on board.
This roadblock is problematic for scientists working to confirm or disprove their hypotheses. It makes it harder for researchers working to ensure a drug will make it through the process and keeps timelines long and expensive. This has a knock-on effect for research around rare disease types specifically. Financial cost and ROI remain big factors for pharmaceutical companies in determining what research to pursue, so rare diseases are often left behind in favour of higher return opportunities.
However, AI is proving to be a valuable resource in tackling this problem. With the support of supercomputers, deep learning mechanisms can now analyse and interrogate enormous amounts of research to form ‘known facts’. Here the technology draws on relationships between data sets and makes connections that may not have been made otherwise.
AI can also help researchers to find out more about existing drugs and whether they might work better in other instances. For example, a drug being used for cardiovascular disease might be found to have useful qualities for neurological diseases. AI can help to bridge the gap between these different fields of research, by allowing researchers to access these insights.
We’re already seeing evidence of how AI can positively impact drug discovery. In fact, this has helped our team of scientists at BenevolentBio establish a set of hypotheses for drug candidates for Amyotrophic Lateral Sclerosis (ALS), a form of Motor Neurone Disease. Together with the Sheffield Institute of Translational Neuroscience at Sheffield University, a world authority on ALS, we are now testing some of these hypotheses in the lab.
Making the most of the data
AI’s potential does not just stop at the R&D stage, however. Opening up additional insights for researchers can not only enhance their understanding, but can also help to improve aspects of the clinical trial stage. This information can better predict how certain patients will respond to treatments and can more readily identify any potential side effects. It can also help clinicians to identify patterns in placebos more effectively and act with more confidence when administering treatments.
There’s no doubt the AI revolution has begun in the pharma industry. The fundamental tipping point will come, however, as we continue to unlock even more untapped data in the industry.
Patrick Keohane is chief medical officer at BenevolentBio
