Latest issue · Episode notes · Guest: Henrik Nakskov
You can't see the data. But you can see what it does.
Henrik Nakskov — CEO of Novyra Life Sciences and founder of CIMS Consult — on borrowing wartime supply-chain theory for clinical trials, the audit that found a changed data point with no audit log, and why AI needs a fence, not a leash.
Henrik is a Danish medical chemist who has spent more than 20 years in clinical drug development — a decade at Novo Nordisk in haematology data management, senior roles at NNIT, and biometrics leadership at Symphogen. Today that experience runs through three companies: CIMS Consult, his life sciences consultancy; Novyra Life Sciences, where he is CEO and Clinical Information Management Lead; and Safe Implant Technology, the research company he leads. The thread through all of them is the same: he wants to know where data comes from, how it moves, and whether it can be trusted when the decisions on the other end affect patients.
AI is like a horse. It wants to serve you, and it wants to make you happy. You can actually flatter an AI machine.
Data has a supply chain
At Novo Nordisk, Henrik's team borrowed the supply-chain management theory developed during the Second World War and applied just-in-time thinking to trial data: data logistics, information logistics, information supply-chain management. Last patient out to database release went from the usual eight to twelve weeks to a matter of days, and the FDA later used the case to teach people what data intelligence actually is. In an ICH stroke study, his team could pinpoint when elderly patients had their intracerebral haemorrhage — because a stroke is not just a stroke: whether it was a bleed or a clot changes everything that follows.
The audit that found a changed data point
His sharpest story comes from an audit of a sponsor exchanging SDTM data over a "secure" FTP site. Inside the CDMS, everything was validated and beautiful. But the exported XPT file could be opened in SAS, changed from male to female, and saved — with no audit log and no change control. Fifteen administrators had access to that folder, with no record of who was in it at what point. His point: compliance doesn't travel with the file. When data leaves a validated system through a channel with no audit trail, you have a hole — and the people now paying to find those holes are investors, who routinely hire people like Henrik to ask one question before an acquisition: can we trust that data?
AI needs a fence, not a leash
Henrik's position on AI is practical rather than starry-eyed. Every deployment starts with a framework: you tell the engine where the boundaries are, then you build checkpoints around it and test it continuously to keep it inside. He is candid about why — AI models want to please. They can be flattered, and they will agree with you. "You have to know potentially the outcome before you ask the question," he says. An AI machine cannot ask questions of the future; it works from references, not invention. Used correctly, AI can speed clinical quality work enormously. Used incorrectly — without real knowledge of your own data — "you can make a devastating tragedy out of your company."
Coaching, not policing
When Henrik runs audits, he opens with a promise: "I'm not here to slaughter you. I'm here to find opportunities for improvement." Quality improves, he argues, when sponsors and sites are coached to understand standards rather than simply policed for mistakes. His dream is continuous, risk-based monitoring that replaces old-school periodic checks — oversight that catches issues early and makes trials both faster and better quality.
Henrik Nakskov is a Danish medical chemist and life sciences specialist with more than 20 years in clinical drug development. He runs CIMS Consult, is CEO and Clinical Information Management Lead at Novyra Life Sciences, and CEO of research company Safe Implant Technology.
Loading preview…