Henning Kuich
May 13, 2024
3 minutes

Metadata Automation is Failing: A Paradigm Shift is Coming

Existing technology, especially the promises around metadata-driven automation (MDA) and metadata repositories (MDRs), will not adequately address the fundamental problems of the industry. They promise to eventually automate 60-90% of all derivations and TLFs (depending on the complexity of the study). However, the remainder is deemed too study-specific or innovative and so is dismissed as a target for efficiency and standardized quality control. 

Ironically, study-specific derivations represent what is most important: derivations in the domains that are important for decision making in regulatory authorities and internal prioritization as well as medical progress itself. By their definition, standards will never adequately capture medical progress and innovation, as they necessarily fall behind the advance of science itself: new methods, new data being collected, and new analyses to prove safety and efficacy. 

Furthermore, MDA is a misnomer in itself. Automation, in this case, is merely a library of pre-programmed code under a different name - a common practice in software and analysis development in any other industry; but nowhere does it claim to automate the bulk of the work: it only automates exactly that for which it was written. Pre-defining everything not only  takes great amounts of time and resources,  it is also impossible - so much so that the industry has become frustrated and early adopters actively seek alternatives after years and millions of investment with failed ROIs. 

Just think about the fundamentals: What is the MDR/MDA endgame? That every company has a store of pre-written code that covers all possible cases of analyses, data integrity issues, and an ever changing environment and requirements? How will we create and store analyses for ideas that do not even exist yet?

Scientific progress will continue to evolve and be ahead of standardization and MDRs, forever and always. All quality control, review, root cause analyses, and resolution will remain inefficient and manual indefinitely if we just bet on MDR/MDA. A fundamental paradigm shift is required to reset the focus on confident, science-based decision making and the processes of constant innovation and progress itself.

Verisian’s products put the focus back on where it belongs: re-using and creating analysis code that is fully traceable and independent of processes like MDA and MDR systems. It allows the seamless exploration of what data analysis is all about: the interaction of data with code that transforms it. When there is full transparency on both of these dimensions, quality control, review, root cause analysis and issue resolution can focus on what is important: scientific insight based on unparalleled quality and resulting confidence.

Explore Further