What is the current state of digitalization in the laboratory? What role do networking, the Internet of Things and big data play in this setting?
Lenk: This is best described by our "five-tiered approach to digital transformation". Tier One refers to sensors that record short-latency and high-frequency laboratory processes. Tier Two is a sensor network. This means that the sensors must be able to deliver data via wireless or wired connection. Tier Three pertains to how this data is stored in a database structure even for long periods of time and is called the "data lake".
Tier Four refers to the "structured data lake". This turns the database structure into a kind of network. Terms and measurement data are thus put into correlation. This gives data context. And finally, Tier Five means making the data human-interpretable. Researchers seek answers to questions such as "Is this a great antibiotic candidate?" or "How do I change culture media for optimized growth?" If you capture data with three dimensions, it often makes it difficult for humans to understand because you can no longer display it in a diagram. Neural networks can process this data. It can be visualized with tools such as mixed reality headsets, voice output, or assistive technologies.
I would say that laboratory digitalization presently ranges somewhere between Tier One and Tier Four, with some applications reaching Tier Five.
What’s next for this area?
Lenk: So far, point solutions have worked well to improve processes and data management, though interfaces are slightly more problematic. For example, when a pharmaceutical company moves a project from development to production, the process often has to be revamped at the pilot or production scale.
While you have the specific formula composition, there is no information about the form factor of the lab reactor, the power input, or the stirring speed. The biggest challenge is to create interfaces to allow all data transfer. To ensure connectivity and a network approach, standards are being created including the OPC-UA initiative by SPECTARIS e.V., or the SiLA Consortium (Standardization in Lab Automation) in Switzerland.
Connectivity also entails difficult adjustments. It is comparatively easy to map the actual workflow analysis and create requirement specifications for the target status. Many believe adjustments can be easily made during ongoing lab operations, clearly underestimating the difficulty of the implementation process. Yet many devices are incompatible because of the lack of standards. Each application requires individual driver programming. Joint standards and interfaces would make transition phases more effective, easier, and faster. Laboratory employees not only have to focus on their manual work tasks, but they must also determine whether the devices are performing the required tasks and produce quality test results. This requires a shift in mindset, and we are still in the beginning stages of our journey.