In this post, we’ll explore how using mobile, voice, and other technologies can help automate lab operations beyond traditional automation.
Along the Digital Transformation journey, it’s not uncommon for lab leaders to evaluate automation technology such as robotics. In this article, however, we will explore a different aspect of automation, looking at how automation can be applied to research & development operations without the assistance of robots. Robotic automation offers several benefits (for more on that, here’s a helpful blog post from HighRes Biosolutions), but sometimes organizations run into issues with space or budget for said systems. Further, some tasks are much better handled by humans, as it can be more efficient for smaller experiments, or dexterity in handling labware.
Automation without robots consists of a manual process augmented by technology to achieve a scientific objective, such as the acquisition, integration, and/or analysis of scientific data.
Concrete examples of scientific workflow automation include:
Adding barcodes to sample containers and storage (freezer, shelves, etc)
Replacing manual transfer of data with integration between ELN and instruments
Guided process execution through a virtual assistant rather than relying on reading instructions
We have many more examples of this throughout our blog and use case whitepapers (links at the end of thispost), but let’s review some of the benefits of automation.
Reduction in Process Variability
As any QC specialist will share, it’s a lot easier to say that you can create reproducible processes than to actually put one in practice. In fact, one study estimates that of the $56 billion spent on pre-clinical research each year, about half of that is not reproducible. When building any type of scientific process (SOP, protocol, method, checklist, etc), it’s impossible to account for all of the deviations and variations that can arise. For example, we recently saw an SOP that had this step for samples containing varying amounts of a powder:
Place the collected samples in a rack in order
The instruction is simple, but what order does it mean? The order in which they were collected? The order in which they came from the oven? Order by their masses? Not only could there be confusion around this step, but as a static document, it could take some time before that confusion is recognized and addressed, thus having downstream implications to the data and product generated in this experiment.
If the lab is using tech to help automate manual processes, like virtual assistants or AR glasses, simple variations as described above can be addressed immediately. As soon as this sticking point is recognized, the proper edits can be made in the software and communicated to the technician following the SOP. Simple edits compound over time, with the effect of creating a more standardized process without necessarily overburdening the person performing the experiment.
Scientists and engineers are tasked with capturing multiple data points, measurements, and observations several times throughout a week. They must have laser-like focus as they perform their work; any misstep can result in bad data, lost time, and wasted materials.
Using the same SOP mentioned in the previous example, here is a list of sample information that the scientist is responsible for capturing in their LIMS by the end of the experiment:
Collection tube ID
Number, names and bottom ID of aliquot child tubes produced
Date and Time of processing
Manual Processing or Automated
Date and Time of Freezing
Some of these are fairly straightforward (date/time), some are more detailed (process discrepancies). And if this list was the only responsibility of the scientist for this experiment, it would not seem such a big ask. But that’s not the case: they must conduct the experiment, transfer the data from the computer to the LIMS, and log observations in a separate report. Outside of any negative impact to the data and product, there is an accompanying and inevitable strain for the scientist.
With automation, much of the experimental and sample metadata can be captured automatically. Integrations can help cut down on the manual entry and related errors into LIMS; audit trails are capable of capturing details like operator name, date/time, and even discrepancies; barcoding reduces the error rate to 1 error in every 15,000 characters.
Automated data collection reduces the 5% human transcription rate to 0%, and boosts traceability of the data collected. This has both short-term and long-term impacts on the lab. In the short-term, the scientist is less focused on menial data collection and with time saved, can increase experimental output. Years later, should a colleague want to run the same experiment, a simple search within their scientific software will yield the results they’re looking for, saving time, money, and effort.
When most people hear “automation”, they immediately associate it with productivity and efficiency metrics. That’s certainly true. Here are several short examples and metrics of technology helping scientists limit the human interaction with experiments and data:
Speech Recognition: it’s estimated that using speech recognition is three times faster than human keyboarding. Not only is it faster, but because speech-to-text doesn’t require the scientist to use their hands, there’s no need to pause in between work to jot down notes, further speeding the process. Medical physicians and other professions often use speech-to-text tools; why shouldn’t the scientist experience the same benefits?
Lab Asset Integration: by integrating scientific instruments and software, scientists are no longer responsible for making ELN data entry their end-of-day activity. In one such case study we ran, we were able to save one process development group 20 minutes per day per scientist through an integrated lab environment. Although this is only a little bit every day, over a month it adds up to over 150 hours saved!
Barcode Scanning: barcodes are really useful in managing a plethora of similar scientific assets (e.g. mice, samples, reagents, etc). Once an accurate list of all assetshas been created, scientists can use barcode scanning to quickly get a sense of how something was used, how it was kept, and any related information.
OCR: Optical Character Recognition, or OCR, is a technology that holds promise to one-day, be able to accurately read handwritten text. For printed text, OCR has an accuracy rate of 99%, potentially saving scientists time from having to copy printed text into a digital format.
Mobile Technology: the emergence of tablets and smartphones in the lab environment, along with accompanying mobile versions of scientific software, has enabled on-demand information, ensuring the scientist doesn’t need to go to his or her desk to find what they’re looking for.
Where to Get Started (Conclusion)
According to one LabVoice customer, each day delayed in vaccine production results in up to $1 million in lost revenue. By using technology to automate workflows beyond robotics, the digital-minded lab is able to make micro improvements that have serious implications down the line.
Interested in beginning to automate some of your operations? We suggest starting with routine, rote tasks that are done on a daily basis. Examples of this include weighing out materials, counting cells, performing animal birth and health checks, and sample location lookup. Why this type of work? Surprisingly, it’s less to do with the immediate benefit of time saved. Rather, we have found that introducing automation technology in this method offers a low-risk method of change management, helping users realize the potential benefit while not disrupting too much in their day-to-day.
Have a process or job you’d like to make faster? Want to improve the quality of the data you’re collecting? Reach out to us (firstname.lastname@example.org), and we’ll show you how it can be done!