Council Post: Digital Science And The Lab Of The Future
Kevin Vigilante M.D., MPH. Chief Medical Officer at Booz Allen Hamilton.
While many industries are being rapidly transformed by digital technologies, research laboratories, like the academic cultures that often permeate them, have been slow to change. They’re plagued by paper-intensive processes, variations in human workflows, fluctuation in experimental conditions, inconsistent equipment calibration and differences in quality control.
Often, laptop and desktop-based processing, spreadsheets and emails constitute the primary technology infrastructure embedded in the costly physical infrastructure. This creates silos that inhibit collaboration and scientific synergies—even in the same laboratory. These conditions also undermine the reproducibility of scientific experiments. In fact, one study found that 70% of investigators had been unable to reproduce others’ research, and 50% couldn’t reproduce their own. The digital laboratory of the future will change this.
The lab of the future is likely to be a data-generating machine that relies heavily on an intelligent laboratory instrument management system (ILIMS)—the digital extension of the traditional laboratory information management system (LIMS). The ILIMS ecosystem harnesses the increasingly prevalent connected instruments and devices in these labs (i.e., the scientific Internet of Things, or SIoT) and brings data, computation and algorithms together for scientific research at scale and speed.
An ILIMS enables continuous automated monitoring of instruments. And if calibration drifts, producing measurement variations in temperature, pH or other parameters, the system can issue automated alerts. Edge AI technology can enable algorithm deployment directly on these devices, allowing advanced analytics at the source of data generation.
An ILIMS and edge AI enable federated model learning while minimizing data movement. In this approach, an AI model is iteratively trained via collaborative learning across devices without exchanging the local training data. This approach to model training keeps data in place on each device rather than requiring cumbersome data aggregation. This also improves data security and reduces privacy concerns. Deploying predictive maintenance AI models to the edge could, for example, decrease downtime and extend the life of expensive laboratory infrastructure, such as high-performance computers.
One of the greatest contributions of an ILIMS system is the automation of knowledge management. Rather than recording experiments in paper notebooks or computers, an investigator can use an electronic notebook that plugs into the network through the ILIMS—automatically connecting research protocols, methodologies and data to a shared or private cloud environment. This allows more efficient data capture and sharing and enables connection to metadata relevant to inventory, instrument integrity and other key parameters. This enhanced documentation of research protocols promotes reproducibility by other teams.
The lab of the longer-term future will likely be much smaller and more distributed. Emerging “lab on a chip” (LoC) technologies emulate the principles of silicon chips to process information. However, instead of etched wiring on the chip, the chip regions and sensors are connected by micro-channels that transport and deliver pico-liter fluid droplets (1 trillionth of a liter), such as blood, for analysis. Constructed by advanced 3-D printing methods, these credit-card-sized chips may be used for biochemical testing, genomics, proteomics and other relevant assays. Similar methods have been developed to grow cells in a 3-D environment to create “organs on a chip.”
LoC technologies could radically reduce capital infrastructure requirements, test time and turnaround, cost per test and required fluid volumes while moving testing closer to the point of care. Organs on a chip can provide insights into organ-specific drug effects, reduce reliance on animal testing and accelerate drug discovery.
Taken together, the rise of ILIMs, SIoT and innovations like labs and organs on a chip is expected to transform legacy laboratories into data factories. The massive troves of disparate and multimodal data unlocked by these labs of the future create the data substrate necessary to catalyze the full power of AI. In turn, AI can allow investigators to identify correlations and patterns that wouldn’t have otherwise been observable or even imagined. This helps generate new hypotheses and discoveries and accelerate science.
However, extracting value from all this data will require next-generation data architectures and paradigms—including cloud-based low schema data platforms (like data lakes) that can accommodate and integrate massive and disparate datasets. It will also require sophisticated machine learning operations (MLOps), a set of practices that allows for standardized collaboration between data science and engineering teams, streamlining the ML development, deployment, monitoring and maintenance life cycle. MLOps best practices enable production-grade scaling and reliability of these models. This helps enhance the traceability and reproducibility of results.
It also will require enhanced cybersecurity to protect the research endeavor from industrial and nation-state competitors and cybercriminals. To combat these threats, future labs will need to increase budgets for IT and cybersecurity, embrace a zero-trust mindset under which users “assume a breach; never trust, always verify; and allow only least-privileged access based on contextual factors” and involve scientists in decisions about designs for cybersecurity solutions.
Finally, achieving the lab of the future will require a different workforce. Many basic scientists are skilled in laboratory and research methods but not in the quantitative and computational skills that are required to analyze the massive amounts of data that digital science will generate. This will require closer collaboration with data scientists and the melding of data science and life science training for those in earlier career stages.
The Covid-19 pandemic underscored the importance of our laboratory infrastructure and network and exposed vulnerabilities in data gathering, sharing and analytics. This applies to academic and commercial research, public health and clinical laboratories. Enhancing the capabilities, speed, reliability and collaboration of these laboratories—individually and collectively—is an important step to improve future public health responses, scientific progress and our ability to compete for technical and scientific dominance on the global stage. We won’t get there with incremental modifications of legacy laboratory equipment and processes. It requires a robust commitment to the laboratory of the future and the advanced digital science strategies that make it possible.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?