Andy York on AI, Data Science & Why 80% Accuracy Isn’t Good Enough | Pharma Prescribed
Adam Walker sits down with Andy York — statistician, clinical data science leader, and former VP of Clinical Data Science at Novo Nordisk — for a masterclass in where pharma's data revolution has been, where it's heading, and why the AI hype needs a reality check.
With over 30 years spanning Roche, Covance, and Novo Nordisk, Andy now advises Verisian, a company building next-generation AI and traceability tools for clinical reporting. He brings rare historical depth and hard-won perspective to one of the industry's most pressing questions: can we actually trust AI with patient safety?
In this episode:
Why 80% AI accuracy equates to a process out of control — and what the Six Sigma reality check means for drug development
The journey from paper CRFs and carbon copies to EDC, CDISC standards, and AI-assisted programming
SAS vs R vs Python — what the shift to open source really means for regulatory submissions
How guardrails, not raw AI, are the key to hitting 95%+ accuracy in clinical reporting
The future of statistical programming: from code developers to validators
Why pharma's AI revolution will realistically take a decade — and what to do in the meantime
Career advice for the next generation entering clinical data science
Whether you're a statistical programmer, data scientist, biometrician, or simply curious about AI's real role in drug development, this conversation will challenge your assumptions and sharpen your thinking.
🎙️ Pharma Prescribed is hosted by Adam Walker, biometrics consultant and independent voice in clinical research and healthcare leadership.
🔗 Listen on all platforms: pharmaprescribed.com
📧 Connect with Adam: https://www.linkedin.com/in/adam-walker-2327951/
⏱️ CHAPTERS
00:00 Introduction
01:00 Meet Andy York — Career & Mission
02:00 Why Pharma? The Noble Cause Behind the Data
04:30 Lessons From 30 Years — Advice for Newcomers
05:30 From Paper CRFs to EDC: How Data Collection Evolved
07:00 Standards, CDISC, SAS & the Programming Revolution
08:30 AI in Pharma — The Reality Check
10:00 The 80% Problem & Six Sigma Thinking
11:00 Guardrails: How to Make AI Reliable in Clinical Reporting
13:00 Human in the Loop — The FDA's Stance on AI Submissions
15:00 Parallel Streams & Freeing Up Resource Through Automation
17:00 Expanding Drug Indications With AI Efficiency
18:00 Conference Trends — The Move From SAS to R
20:00 Open Source vs Validated Systems — Regulatory Perspectives
23:00 SAS, R, Python — The Right Language for the Right Job
24:30 Future Roles: From Programmer to Data Scientist
27:00 The Next Generation — Gamers, AI Natives & Skills for Tomorrow
29:00 Science Fiction Becoming Science Fact
30:30 Verisian's Roadmap — What's Coming Next
32:30 Andy's Closing Advice for Programmers at Every Stage
🏷️ TAGS
clinical data science, statistical programming, AI in pharma, drug development, clinical trials, SAS vs R, open source pharma, CDISC standards, regulatory submission, FDA AI, Novo Nordisk, Verisian, biometrics, data science careers, pharma podcast, clinical research, Adam Walker, Pharma Prescribed, Andy York, EDC, SDTM, ADaM, PharmaVerse, R consortium, double programming, AI accuracy, patient safety, pharmaceutical industry, clinical data management, real world evidence, pharma careers