Big data has well and truly arrived. But for all its enormous potential, it presents unprecedented challenges to researchers in the field. “Scientists, from research technicians to principal investigators, spend too much time searching through data,” says Zareh Zurabyan, head of eLabNext, a specialist in laboratory digitization. Before long, there will be too much data to manage without help. “Rapidly accelerating data generation has outstripped the rate at which human beings can meaningfully digest,” says eLabNext Founder and CEO Erwin Seinen. “This means artificial intelligence (AI) will become a necessity for biotech, not just a novelty.”
There are many opportunities for machine learning (ML) and other AI-based tools to transform the research process. For example, such algorithms can guide the execution of automated multi-instrument experiments or identify patterns within a research pipeline. When these tools are directly connected to electronic lab notebooks (ELNs), such as the one developed by eLabNext, it is possible to seamlessly integrate AI capabilities into researchers’ daily workflow.
The use of AI in drug development is not yet a matter of course. Deloitte’s 2021 research found that 38% of biopharma companies surveyed use AI every day, although another 31% are exploring the use of such tools.”In many cases, we’ve seen senior management absolutely support the integration of AI into their research, drug discovery and precision medicine,” says Taylor Chartier, founder and CEO of life sciences company Modicus Prime. However, he points out that implementing such capabilities can be a struggle. “Pharmaceutical companies are not software companies.” Effective digitization requires access to external AI-based tools and services that minimize the pain of integrating with existing workflows, he adds.
For labs that are software-naive, there are many benefits to adopting AI tools. And for those that are already digitized, there are strategies to make the most of AI technology. eLabNext complements its ELN platform with two AI add-ons, developed by Modicus Prime and ImmunoMind, a startup specializing in single-cell multi-omics for cell therapy development, to seamlessly meet the data needs of biopharmaceutical laboratories from image analysis. to identify the cell.
How can AI be useful?
For those new to the world of AI, it can be difficult to distinguish the hype from the reality in terms of what the systems can do. “AI can generally solve three kinds of problems,” explains Vadim Nazarov, co-founder and CEO of Immuno Mind. “It can automate small and simple tasks, increase the performance of some complicated tasks or, if you have a lot of data, provide an opportunity to gain insight.”
Before a biopharmaceutical company decides to adopt a new AI-based system, it needs to have a clear idea of the problems it wants to solve and how the tool can streamline or improve existing processes. This puts the onus on AI firms to develop tools with clear and compelling applications. “There are some really incredible AI companies that have access to massive amounts of data,” says Chartier. “But whenever you introduce a new technology, it has to be very relevant to a particular customer.”
For Modicus Prime, this meant developing an AI-based image analysis tool that could be easily trained on specific research problems. Chartier notes that computational analysis and image interpretation are among the most advanced applications of artificial intelligence in the life sciences. Her company’s mpVision software can quickly analyze most categories of biotech imaging data. For example, users can distinguish desired cells from other cell types or cell debris, detect anomalies during drug manufacturing, characterize crystallization processes, or perform rapid quality control of biologicals at any scale—from the lab bench to the production floor.
Designed for more specific immunology applications, ImmunoMind software draws on proteomic, transcriptomic and other data types to help identify T cell subpopulations and characterize their physiological state. This may be particularly important for quality control in areas such as cancer immunotherapy, where donor-derived T-cell subsets are cultured and genetically manipulated to selectively target and kill tumor tissue. “Finding relationships between gene expression and different cell phenotypes is extremely important for the development of cell therapy,” says Nazarov. “These are tasks that simply cannot be solved with traditional statistical methods – only with machine learning algorithms.”
In AI, data is king. The performance of the algorithm strongly depends on the quality of the training data and subsequently the experimental results embedded in it.ImmunoMind addressed the former by building a curated training database based on multi-omics analysis of a vast array of immune cells. The company then provides a user-friendly portal for researchers to use this data to gain insights into their own cells. “We work closely with customers to help them design experiments and quality control measures to eliminate all risks associated with batch effects and bias from non-ideal experiments,” says Nazarov. In contrast, the more general mpVision image analysis framework is trained by the user; Chartier says that just 20 representative snapshots from a particular experimental process can be enough for AI to evaluate future data from the same pipeline.
No add-on requires any formal training in AI or expertise in computational biology, and user-friendly interfaces come standard. Having the right underlying infrastructure for data management – such as a laboratory information management system (LIMS) – is also critical to the effective use of AI tools.Zurayan suggests that companies can further accelerate adoption by having staff focused on the task. “One well-received approach is to lead teams that dedicate time and effort to a strategy to implement new technologies within a specific time frame with very clear goals and milestones,” he says.
For lab staff who don’t fully understand what algorithms do, there’s a natural fear of the unknown indeed, many AI systems have been criticized as “black boxes” that rely on convoluted and obscure processes. However, AI developers can achieve some degree of transparency by explaining the underlying mathematical models and providing audit procedures that allow users to check the machine’s performance. This is particularly important for scientific software intended for use in highly regulated environments such as Good Manufacturing Practice (GMP) facilities. GxP-compliant ELNs such as the eLabNext platform do this by automatically tracking and logging the movement of data within the system and its various accessories.
Source Journal Reference: https://www.nature.com/articles/d42473-022-00089-y
