Pulling ourselves together

As laboratory equipment becomes more interconnected, can the Internet of Things help tackle the ‘reproducibility crisis’? Possibly – but, say Nicolas Paris and Klemen Zupancic, all vendors and scientists need to pull together on this…a common framework really is the only way forward.

The ‘reproducibility crisis’ in the life sciences continues to make headlines, with findings that show well over 50 percent of published studies cannot be reproduced (some experts cite numbers as high as 90 percent). There are many causes. Researchers perform complex tasks with limited grant funding and face ongoing pressure to publish compelling results. Despite these challenges, reproducibility and repeatability are aspects of science that cannot be compromised on. The field needs to have confidence that published results are accurate and able to stand the test of time.

Instead of changing scientists, we advocate for changing the scientific system from within. Scientists are losing valuable time struggling with data management issues, when they could instead be concentrating on meaningful discoveries. Connecting instruments with each other and with the cloud – the so-called Internet of Things (IoT) – can minimise these distractions, removing human error and the disconnect between scientists and study results. In our highly connected world, saturated with personal technologies, this online transition would seem inevitable. Well-designed and well-integrated IoT systems can improve lab efficiency, data reliability and study reproducibility. But to date, the field has been reluctant to adopt the technology and currently lacks the critical mass ‘online’ to begin solving the macro challenges. What needs to change?

[caption id="attachment_54546" align="alignnone" width="620"] Using the IoT could make reproducing scientific research easier.[/caption]

It’s incredible to think that we are only now moving beyond paper notebooks in the laboratory. Most scientists still write protocols by hand and upload data to local servers. Each instrument and each scientist operates independently, and with each step in the workflow a degree of precision is potentially lost. For all its flaws, manual recording does keep scientists in control. For IoT-enabled lab systems to replace them, they need to be equally responsive. The overarching technology has to be robust and reliable – yet sufficiently versatile to allow for the huge scope and complexity of modern life science experiments. Few – if any – lab equipment and service providers can achieve this on their own. We need industry collaboration to combine our areas of expertise into a workable, cohesive system. A system that truly works for scientists. Below are some of the lessons our teams have learned through partnering with scientists as they trial cloud-connected products and services. As an industry we have the rare opportunity to strategically build something greater than any one company can achieve, helping labs around the world transition to cloud-based technologies in a mutually beneficial way. If we don’t work together, a system of closed circuits and technology stand-offs could result, impeding our ability to advance science in a reproducible, verifiable way.

Instead of changing scientists, we advocate for changing the scientific system from within

To compel scientists to adopt a new technology, there has to be a clear value. IoT promises this and more. Imagine if, as scientists pipetted, the data could be used to correct their technique. Bench scientists would no longer need to write down every step in a physical notebook and then type those notes into a computer. The IoT-enabled instrument would take care of data and protocol logging so the scientist could focus on their science. Data stored in the cloud would be available for any researcher to verify or reproduce published experiments – even years later. In this scenario, IoT automates and standardises mundane laboratory tasks, allowing the scientist to focus their energy and resources on the decisions that matter most. Our recent market research suggests early cloud-connected lab instruments and data systems have not catered to scientists’ needs in this way. While there is growing interest in implementing systems such as electronic note books, adoption rates remain low. There is also a trend towards customised systems built in-house, suggesting the existing designs don’t translate well in the lab. This delay and lack of standardisation helps allow the reproducibility crisis to go on.

Few scientists have the time to relearn systems and no one likes to troubleshoot program quirks or deal with technology breakdowns. IoT systems shouldn’t represent a “new way of doing things” – rather, they should enhance what scientists already do. For example, for users who are already comfortable with our pipettes, we created an option to retrofit existing instruments with Bluetooth capabilities. It’s a simple cap that fits over the plunger button for added functionality. Protocol and data are automatically uploaded and can be transferred to other similarly cloud-enabled instruments in the workflow. User-friendliness has also been a vital software design consideration in ELNs. Scientists’ careers are built around specialisation and for the most part that doesn’t include training in bioinformatics. The option exists for advanced programming, but we’ve found only a small population of scientists are capable of, or interested in, writing their own code. For automated programs to succeed, the focus should be on the real human benefits the technology enables, not the advanced features. However complex the innovation is, it needs to simplify and streamline lab processes.

[caption id="attachment_54547" align="alignnone" width="620"] Adding the IoT to a lab can enable researchers to have their results noted down for them as they work.[/caption]

When it comes to purchasing decisions, costs are front-of-mind for the majority of labs – certainly those in academia. Having limited budgets was cited by our survey respondents as the number one barrier to entry when it comes to ELN and cloud-connected technologies. If vendors were to take a transparent, collaborative approach, it would be possible to achieve a highly cost-effective transition. For example, open-source software represents the collective skills and innovations of everyone who uses the technology. Scientists thus become both customers and software engineers, feeding into a collective reservoir of knowledge. With this shared responsibility, labs can connect and begin storing data in an organised way, without paying for a premium service. As we embark on this new era of science, we’re aware that an entirely different scenario could unfold. For the system to work, IoT providers must develop programs that are compatible with new and existing software and cloud-connected instruments. The same is true for hardware. If this does not become the norm, the efficiency benefits of cloud-connected technologies will not materialise. Labs around the world need access to a universal data language.It is our belief that no life science company today can offer top-of-the-line products and services across the board. Each has areas of expertise as well as areas where competitor’s products may perform better. As instruments and technologies become connected, we need to ensure that scientists have the right to choose from a full range of vendors. This will be undermined if instrument and/or data management providers decide to pursue closed systems. A closed circuit, or an incompatibility stand-off, would hurt the efficiency of laboratories and the progress of science more broadly.

IoT systems shouldn’t represent a “new way of doing things” – rather, they should enhance what scientists already do

An often-overlooked concern is the possible amplification of economic inequality between labs worldwide. We get excited about the immense potential of cloud-connected instrumentation streamlining lab processes and enabling unprecedented collaboration. But if this is not done in a cost-effective way, we risk forming a separation between “online” modern labs and labs in the developing world that remain offline. Clearly, large social and scientific costs are associated with marginalising any part of science’s diverse global research team. Where possible, simple open source systems should prevail, to keep costs low and accessible to all. We’re on the cusp of a fundamental transformation in how R&D is performed. It’s born out of a need to make science reproducible. With funding agencies such as the National Institutes of Health (NIH) calling for strong measures to incentivise and/or penalise academics whose findings cannot be substantiated, scientists will search for best practices to make that happen. We need to make those tools accessible and intuitive, for a critical mass to change how they program and record experiment data. We believe cloud-connected tools are part of the infrastructure needed to solve these systemic issues.

What will a life science laboratory look like in ten years’ time? Conceptually, the same – at least on the outside. What will be most different is what’s unseen: the information flowing between devices, through the clouds and amongst operators, producing experiments that can be replicated time and time again. The instruments will take care of the recording, allowing the scientist to focus on the big picture and the interpretation of results. We look forward to continued discussion and collaboration to ensure viable, open systems get off the ground, to truly advance science on a global scale.

 

Authors:

Nicolas Paris, CEO of Gilson is an expert in sustainable development of businesses in the lab instrumentation industry.

Klemen Zupancic, Ph.D., CEO sciNote, leads a team of life scientists and software engineers, developing an open source electronic lab notebook called sciNote.