In this exclusive article, Dr Peter Crane, corporate strategy manager at Synthace, explains why augmented and assistive technologies could enhance the scientist and provide the first step towards a data science-led future for biopharma.
Like many other businesses, biopharmaceutical companies have begun to tentatively invest into advanced analytics such as machine and deep learning. Despite this, some senior industry figures are now questioning if we are ready for these techniques, if we have the depth and breadth of data necessary to truly leverage them.
The challenges to handling biopharma data are complex
As was stated by Google in 2018, applying advanced analytics to a research lab environment necessitates the collection of a greater amount of data (higher content and with context). Investing into professional ‘data wranglers’ may be a short-term solution, but this approach is akin to treating the symptom rather than the cause. This will become especially evident as the impact of the data sciences grows, with companies attempting to wrangle data across geographies, sites, functions, cloud providers, instruments, legacy servers etc.
Working in a compatible way with data science can quickly become prohibitively burdensome if experiments are conducted using traditional, and often manual, ways of working. Data is often recorded in a subjective and observational manner, which can lead to errors and inconsistencies whilst also being a large timesink for scientists.
Rather than contributing to and increasing the burden of siloed and machine learning-incompatible research data, researchers should be empowered to work in ways that are forward compatible with the envisaged future needs of data scientists, as well as incentivized to adopt FAIR principles.
One way to enable this is to focus efforts towards the adoption of foundational informatics tools, which directly interact with their environment through technologies such as the Internet of Things (IoT) or lab automation. When coupled with a movement towards statistical ways of working such as design of experiments (DoE – which inherently produces rich structured data), we can begin to realize the promised benefits of advanced analytics in the lab environment.
Scientists should be at the heart of the digital transformation in the lab
In industries like the financial services, purely digital tools are having transformative effects, such as large data sets being collected in homogeneous electronic formats, before being automatically parsed and interrogated for operational or market insights (e.g. with robotic process automation (RPA)). In biological research and development (R&D) we are still often dealing with heterogeneous ‘wet lab’ experiments. Technologies that straddle both the digital and lab domains will become increasingly important in delivering the Lab of the Future.
These technologies will essentially fulfill a role akin to the human researcher: collecting data, analyzing it and then using this knowledge to formulate and run further experiments.
However, rather than simply replacing the researcher, in order to deliver tangible benefits quickly, we should focus on technologies that assist or augment the human capital of biopharmaceutical companies and empowers them to focus on value-adding innovative tasks.
Computer Aided Biologists will pioneer new research
Technologies that empower scientists to work in digitally compatible ways can be classed as assistive or augmented intelligence, as they put the scientist at the heart of the digital transformation, empowering them in a non-autonomous way.
Within the wet lab, a technology that enables scientists to automate the collection, structuring, collation and annotation of diverse data flows may be viewed as an assistive technology. These technologies empower researchers to focus on more productive tasks and deliver benefits such as increased throughput and reduced cost.
When using an assistive technology, the researcher is still very much in full control. The natural progression of these technologies is to combine these assistive elements with faster and smarter feedback loops, and thus move towards augmented intelligence. In this partially closed loop the system is able to navigate the defined experimental design space, with the guidance of the expert user, and make in-process experimental changes in response to real time analytics. This is also known as Computer Aided Biology.
This movement from manual, to assistance, through augmentation and finally potentially autonomous fully closed loop R&D draws analogies to the different levels of autonomous vehicles beginning to enter the market.
Assistive automation empowers the scientist
A basic example of an assistive technology in action was demonstrated by a leading technology consultancy who, by using a digital automation and informatics platform from Synthace, were able to reduce operator hands-on time by 81% and increase throughput by 50% for a key analytical method (qPCR). If combined with environmental monitoring via IoT, the system would enable fast execution of a common laboratory assay, with environmental and historical sample traceability contexts – perfect for downstream machine learning. Vendors working on process analytical technology integrated into bioreactors, as well as R&D intensive companies working on ‘closed loop’ automated high content screening are also perhaps closer to augmented intelligence, albeit only for discrete use cases.
Building a digitally empowered R&D organisation doesn’t happen overnight and cannot be delivered simply by hiring a silo of data scientists or wranglers. Instead companies need to look to break down or prevent the formation of cultural silos and look to curate an internally common language that bridges science and computing.
Requirements around the quality, quantity and diversity of data needed for advanced analytics should be agreed across the organisation. To implement such a transformation, we need to change the way we conduct biological R&D. Moving towards ways of working that are built around the high-throughput capture of high-content and contextualized data, and to do this without burdening our teams we need technological solutions. The approach laid out here of assistive and augmenting technologies provides a tangible first step towards a data science future for biopharma.
About the author:
Dr Peter Crane is the corporate strategy manager at Synthace. At Synthace he is responsible for new market exploration/expansion and thought leadership around the ab of the Future. In 2018, he co-authored the comprehensive Synthace whitepaper “Computer Aided Biology: Delivering 21st Century Biotechnology”. He has a DPhil in chemical biology from Oxford University & a background in life sciences venture capital.