Patient-centric predictive modeling: AI-ML
Among the billions of individual data points and genetic interactions, there exists an infinite number of possible pathways we could investigate. Given the complexity of this data, we need more advanced tools to help us analyse it at scale. That’s where AI comes in, to help us separate the relevant from the interesting.
Dr. Fiona Behan, a Scientific Leader of Functional Genomics at GSK and former cancer researcher at the Wellcome Sanger Institute, sums up the promise of these tools perfectly. “Before, we knew what genes were, but not what they were doing,” Behan says. “It could take scientists decades focusing on a single gene to figure out the function. But by combining functional genomics experimental approaches with AI and ML, we can often figure that out in a couple of months.” She adds, “The reason I do what I do [in the field of functional genomics] is to transform patients’ lives.”
With our in-house AI hub, we’re using ML to unlock the potential of complex genetic data with never-before-seen levels of speed, precision, and scale. Every model we build, run, or test deepens our understanding, and creates a compounding, near-infinite feedback loop of experimental possibilities. Ultimately, the use of AI improves predictions, viability, and the speed at which we can bring investigational medicines and vaccines from the lab to the clinic, and most importantly to patients.
At GSK, we believe that the predictive power of AI and genetics holds the key to improving research and development. Kim Branson, Global Head of Artificial Intelligence and Machine Learning at GSK, also looks to the past. “We have a history of everything that [humans have] experienced in our genes…and we now have the tech to read the language of the history in our genome, make comparisons, and test hypotheses.”
