by Jonathan Gross, Group Chief Product Officer, Titian Software & Labguru
For any lab environment, AI adoption is a when, not an if. However, AI is not going to be the “be all and end all,” to creating fully efficient and effective lab environments.
Fully embracing lab digital transformation will.
Digital transformation, of course, varies by lab. Some plan to leverage robotics and smart automation to create 24/7 lab operations; others want to accelerate analysis by leveraging the ever-increasing volumes of data their lab is generating.
According to a recent survey of more than 150 lab professionals across labs of all sizes—from small start-up to big pharma, AI adoption is just one of the activities necessary to transform lab operations, from all manual or semi-manual processes to mostly automated or completely automated labs—achieving digital transformation.
When asked about their current digital lab set up, less than 20 percent said their labs were fully digitized, and more than one third still depend on manual processes.
Embracing next-generation technologies is what will bring next-generation science to the forefront, accelerating scientific discovery while increasing efficiency, reducing waste—both materials and time, and allowing labs to scale without extraordinary investment.
Survey Background
In January of this year, more than 150 life science professionals completed the Titian Software-Labguru survey, providing critical insights into planned innovations and trends in the quest to achieve full digital lab operations. Labs represented included pharma, startups, chemical companies, Greentech, and government.
The survey respondents:
- 26% from large pharma/biotech (>5,000 employees)
- 12% from mid-size pharma/biotech (500-5,00 employees)
- 29% from small (startup) pharma/biotech (<500 employees)
- 4% from CRO (Contract Research Organization)
- 21% from academic institution
- 8% from other organizations
We will be referencing the survey throughout the article.
Samples—The Heart of the Lab
While hardware, software, personnel, and supplies are critical components of lab operations, without samples, there’s no science. Therefore, sample management needs to be a core lab process—without it, chaos reigns. Sample management comprises three separate components—effective tracking, data integrity and proper storage. If you can’t find it, you can’t experiment with it. Meanwhile, poor data integrity may cause inconsistent results, significantly impacting research reproducibility. Cluttered storage means it takes much longer to find your samples, and if the clutter affects temperature, the samples may degrade to the point of not being usable. And if you have too many expired reagents in the refrigerator, you’re simply wasting a lot of energy.
Proper sample management requires an investment in a simple sample management system that supports complex processes behind the scenes. The solution needs audit trails for compliance, lineage tracking to ensure repeatability, and simple barcode and storage management to make it easy to find the samples and ensure they are stored properly. Of course, when assigning the barcode, critical data should be noted such as the expiry date, concentrations, and volumes.
The Challenge of Data Management
If the sample is the heart, data is the circulatory system of every lab —collecting it and analyzing it, moving through the entire experiment to support the quest for amazing pharma, life science, biotech, and manufacturing breakthroughs.
Managing that data is both a challenge and an opportunity for innovation, both in the data management space itself and the drive toward AI. Over half of the survey respondents said that data overload / data management would be most likely to drive change in the lab.
Data management is a challenge for a variety of reasons. One is because of the lack of orchestration of lab software and hardware. Data is collected in silos, with a variety of formats. That data must be combined, collated, and standardized before analysis can even begin. Orchestrating that data is the first step—whether with AI or simple integration tools.
Beyond Ordering—Keeping Up with Inventory Management
So, while we emphasize data is at the heart of the lab, the “body” that supplies that lab is inventory—the reagents, the assays, the samples, the molecules, and so on.
Inventory management is still one of the more manual lab processes. Storage data is often found across spreadsheets, and if someone forgets to enter the fact that the lab is down to the last bottle of a critical reagent, when it’s time for the experiment after it’s used up…
Many technologies exist to streamline lab management, everything from mobile apps that allow you to enter inventory data on the fly—integrated with automated shopping lists and ordering infrastructure, of course—to lab robots that physically hand you the necessary chemicals and always know where everything is stored and automatically order when their supplies are getting low.
Inventory management systems are critical in the quest for digital transformation. No one has to check every freezer to find that last bottle of reagent that may be long expired.
Embracing Automation
Lab automation is crucial for research and development in fields like pharmaceuticals, biotechnology, and diagnostics where speed, accuracy, and scalability are essential.
Many hardware and software solutions providers have joined forces, delivering end-to-end automation solutions for labs. They include everything from automated instruments that analyze one sample right after another and feed the data directly into lab management software to robotic systems that automate sample preparation, pipetting, and even cell counting.
While the initial investment in automation may pose a budgetary challenge, it will improve reproducibility, accuracy, and consistency while reducing manual tasks and streamlining data collection and analysis.
Many labs surveyed said that automating processes was a higher priority than AI adoption itself, no matter the size of the lab— from startups to big pharma. Only 15 percent of labs are fully digitized; 50 percent still rely on manual processes.
Tying it all Together with AI
AI is going to be the workhorse bringing together data, sample, data, and inventory management and automation to achieve previously impossible discoveries and advancements. It’s going to improve sample scheduling and tracking; reduce human error; enable predictive insights; speed data interpretation; generate hypotheses; and even design experiments. It will also make it easier to manage the massive data volumes generated by instruments, experiments, sample analyses, and more.
Another data challenge that AI is expected to address is managing data across the range of data-generating lab tech: automation, acquisition, storage, equipment, inventory, etc.
AI is going to replace some of the day-to-day manual tasks, like scheduling lab equipment. People will be able to put in their requests and based on the “before and after” experiments, AI will be able to slot them into the most efficient time windows. Also, before the massive data sets can even be analyzed, they must be standardized. AI can take those disparate data sets and standardize them, making it much easier for the “next step” of AI to analyze them.
One aspect of AI that generally doesn’t get considered is its ability to increase the lifespan of common laboratory equipment, reducing CAPEX. Many manufacturers are now integrating AI tools as part of preventative maintenance, notifying lab personnel when equipment needs to be cleaned, when a part needs to be replaced, or a “standing” reagent needs to be refilled.
AI is also going to be able to “crunch” data at speeds and volumes people just can’t achieve. Traditional drug discovery methodologies require hundreds (if not thousands) of people hours testing different combinations of molecules. AI can “test” the various combinations and provide faster insights into drug discovery.
While long-range thinking expects the main benefit of AI to be analyzing data and generating insights that “humans” just can’t “see,” the initial benefit of AI is eliminating many manual processes, allowing scientists more time to think and focus on more critical aspects of their research.
AI and Compliance
AI’s relationship to compliance is definitely a gray area right now. AI will most certainly improve data documentation and integration. However, the compliance of AI within the lab itself is going to be a significant factor.
The regulatory environment varies by both region/country and industry, so integrating AI into lab operations is going to require consultation with the compliance team. Scientific oversight also needs to be taken into consideration—can the experiments be reproduced without input from AI? Findability, Accessibility, Interoperability, and Re-usability - FAIR data principles—need to be a given within any AI adoption.
In addition, the basics still hold—integrating AI into lab operations requires consultation with not only lab management but also the IT team and anyone else with a vested interest in how your lab operates.
Ethics should also play a role—what is appropriate for AI to do?
The role of AI is to supplement, not replace, human ingenuity. The more a lab relies on AI, we must consider the long-term implications. Dependency on AI can grow to levels where we are no longer supervising the technology, but it is supervising us, which could be an issue when considering new experimental approaches. We need to be in the driver’s seat, not sitting in the back row of the AI bus.
Not Quite Full Speed Ahead with AI
The promise of AI is still just a promise. While almost half of the survey respondents plan to implement AI within the next two years, 25 percent expect to implement AI more than five years down the road, as they have a clear understanding, they need to get their lab in order before they take more advanced steps.
Whether a lab is on a two-year timeline or a five-plus-year timeline, integrating AI is going to be a critical part of scientific research and the quest to remain not only competitive but also simply viable. Even if labs aren’t ready to become completely AI-focused, they need to start “acting now” to make sure they are on the right track, automating where they can, eliminating manual processes, and supplementing their activities with AI when appropriate.
About the author
Jonathan Gross is Chief Product Officer of Titian, which offers the Mosaic Sample Management platform and Labguru, a transformative lab operations platform. He previously was founder and CTO of Labguru, where he spearheaded the development of solutions that reimagine how research is documented, shared, and analyzed.