The Case for Automation
Relationship between Humans, Processes and Automated Equipment
Process and system automation has great appeal; it potentially reduces the need for manual intervention as well as increasing throughput with processes running 24-7. Most lab processes are geared around the regular 9-5 working day – although many scientists work extra hours to ensure experiments are executed properly.
Process automation and systems refinement have both been used in many other industries over the years – notably in the car industry, where automation revolutionized their ability to deliver new vehicles and improve quality.
In comparison, science and pharmaceutical research and development (R&D) has higher barriers and complexities. It is quite normal to see automation used in the manufacturing plants of pharma and biotechs, but when we look further back into R&D areas – where drugs are discovered, tested and optimized – then the processes are far trickier to automate.
Nevertheless, some parts of this R&D process can already be automated. There has been a real change in the way laboratories have been organized and new technology has been developed to help with the mundane and laborious tasks.
Robots have been around for many years in the laboratory, and used heavily in the high throughput (HTS) screening areas. But we have also seen new instruments arrive on the market that provide elements of “robotics” too. Many of the newest analytical instruments have some form of automated sample handling and injection controls – removing the need for people to directly use a syringe.
However, many of these analytical processes also involve important process steps before and after the “analysis,” including sample preparation, layout and data reduction. All of this “data” that is produced is also needed in other parts of the process. Just “running the process” in an automated fashion is not enough – the data flow needs to be automated too.
Data around samples may come from different parts of the laboratory or other external organizations. Many moving parts and complexities are hurdles to automation in R&D, but some areas are very amenable. They are characterized by processes that are very consistent from numerous industry studies, have set inputs and outputs, and require little manual intervention.
Impacts on Laboratories
One approach in these areas has been to rethink the layout in the laboratory – for instance, moving to a “workspace” or “work stations” that are geared and organized to promote efficiency. Rather than having the lab with instruments in groups, requiring scientists to move around, all equipment and instruments for a process are in one area.
This workspace approach helps reduce manual intervention, but full automation is the next step. This requires a lot of extra technology to make it work (automated sample scanning and ID, fault handling, reagent handling etc.) so can only be used in specific, well-defined process areas.
All of this automation comes with a data management and data reporting requirement alongside it – adding more complexity.
Process automation in R&D has allowed laboratory staff to perform other functions or run concurrent activity. This has given laboratory managers the opportunity to reassess where their scientists and researchers can provide value, and how employees’ time can be used in other roles. The tricky thing is showing its impact on reducing time to market and the cost of new drugs.
Cloud Computing, IT and Integrated Offerings
One area that has been of great interest in the past 5 years is the cloud, and how it can reduce IT burden and costs.
Many IT and informatics organizations admit that cloud, Software as a Service (SaaS) especially, brings great benefits: reduced total cost of ownership (TCO), better uptime, resilience, performance and global support. There are many examples of this being clearly demonstrated in research organizations, but there is still a hesitation and reluctance in some companies that cling to the belief that their costs are lower than they really are.
This is where costs for running a system are not put under one line item but distributed across other group budgets – a good example being power costs for hardware. This tends to give a false view of what is costs to run a system and blurs the arguments for moving to a cloud SaaS environment. But, when all costs and elements such as better performance and happier researchers and scientists are considered then cloud and SaaS makes perfect sense. What does the combination of instrument workstations, process automation and SaaS really mean?
We are now seeing the emergence of IoT (Internet of Things) SaaS service companies that solve specific problems in the laboratory in areas such as reagent and solution preparation. Here, you need stock reagents and stock checking, recipes for solutions, an instrument to make the solutions following the recipe, checking for conformity and dispensing. All this is already possible when done manually or with a combination of hardware and software, but the difference now is that the instrument is connected to a cloud service that monitors the status of the instrument and can alert the manufacturer to any issues or stock requirements in real time. In this scenario, it is also possible to check that reagents are mixed and solutions prepared to the recipes specification, and notify the service provider and scientist if statistical drift is occurring. All the scientist needs to do is place a bottle to collect the specified solution – everything else is taken care of. This is exactly how aircraft work – when in the air, aircraft are constantly communicating with their manufacturer to provide updates on performance and maintenance requirements, so when it lands parts and technical experts are ready and waiting.
In the research example, all information about that automatically prepared solution such as the audit trail, reagent batch information, reagent ID etc., is automatically included into the scientists’ experiment recording system (ELN or LIMS) via back-end integration of the two systems.
This might seem like a trivial process, but the concept is a perfect IoT example already being used to optimize lab processes today.
Such innovations dramatically reduce the time spent making and checking solution preparations, and greatly improve the repeatability – thus improving the overall quality of the solutions being made. This can be extended to numerous other tedious processes, and will improve both efficiency and productivity of research organizations using cloud and IoT approaches.
It will, of course, require investment, changes in laboratory process and scientists’ mindsets, but sometimes technology can be better than processes that exist now.