Data Science and AI
Julie Huxley-Jones, BSC Genetics, PhD Molecular Genetics and Computational Biology
VP Research Solutions
GlaxoSmithKline, England, United Kingdom
Science moves fast. Technology moves fast. Investment cycles rarely do. Money isn’t unlimited.
In the world of the modern automated and connected laboratory, sustainable data and technologies need to be both nimble and robust, resilient over time to enable lab research to evolve whilst not spiraling in cost.
We want the smart scientist to focus on using her expertise for insights and decision making.
We want to build around her a world of augmented and distributed environments with frictionless interfaces that enables her to easily engage with a dynamic range of equipment as she executes her experiments. Paperless labs work around her, data is automatically documented directly from smart equipment, laboratory automation performs standard tasks, and the first drafts of reports are automatically built using AI-ML enabled authors. Equipment runs 24/7 but she doesn't have to, fully utilising IoT, industry 4.0 and remote-control connections that seamlessly live stream for real time and off-line decisions. No experiment is wasted as all data flows into AI-ML enabled models which inform future upstream experiment designs. Quality is embedded and data integrity is bullet-proof.
However, historically siloed, narrow decisions of functional laboratory leaders across the life sciences industry have inadvertently shaped the current application landscape in what the outside world sees as a relatively niche area. Technology solutions rarely support both R&D and manufacturing laboratory environments or the regulated vs non-regulated spheres. No discovery scientist wants to be locked into GXP for no reason, no manager wants to pay for GXP and non-GXP versions of the same technology. Single-purpose lab applications also frequently require extensive professional services to integrate. Rarely does a solution come off-the-shelf.
If not managed carefully, the research scientist interacts with more products than any other user in any other industry! Her time is too important to train on yet another system.
So, what do we do?
There is a sweet spot between full flexibility to operate in an ever-evolving dynamic way and managing the technical debt for a consistently sustainable laboratory environment.
We will outline strategic approaches to make a data science and AIML enabled laboratory robust, relevant and sustainable.