5 Ideas That Stuck With Us from #ddpeast26
Explore insights from #ddpeast26
1. Evidence is the bottleneck, not the data. More data collection isn't the cheat code. Causal, directional evidence is. Targets with human genetic support succeed at twice the rate. Or, as Nevin Ince (Novo Nordisk) put it for anyone still buying GPUs to fix a data problem: the bottleneck was never the algorithm. It's whether the data can actually tell you something.
2. The hardest part of an eight-year data platform is patience. Birgit Schoeberl's keynote on Novartis's Data42 (eight years, 3,000 trials, ~1M patient lives) was clear about the part nobody puts in the launch deck: it almost died halfway in. The hardest problem turned out not to be engineering. It was surviving long enough to be right. A bracing palate cleanser in a year of "we vibe coded it this weekend and now have 5M ARR" energy.
3. Every early R&D dollar is worth almost $18. Sun-Gou (BridgeBio) shared math on how to think about drug R&D investment: the first $1M in preclinical needs ~$17.9M in revenue to earn its keep, when adjusted for risk and market returns over drug program timeframes. Strong human genetic support does about as much for your program as cutting costs 40% across every single phase.
4. In Pharma, we don’t do it because it is easy, we do it because we thought it was easy. Joanna Kemp (GSK) said this about digital health. The concept is great (monitor your patients more closely for better clinical trial data and outcomes), but the devil is in the detail, and the details aren't pretty. The underlying concept also rang true during Eric Ma's (Moderna) talk about agentic data science, and Nevin's deep dive into digital lab AI systems. Everything is harder than it seemed at first.
5. The human side of AI and a perspective on the future of the Data Science role. Elizabeth Choe (AstraZeneca) expressed her individual views by naming what many feel: offloading the easy tasks packs your day with harder ones, plus the new tax of filtering AI slop from others. She also drew a sharp analogy to accountants. Tax software didn't replace them, but the profession now faces a CPA shortage, partly because firms stopped hiring entry-level talent to grow into senior roles. She sees the same risk for data scientists: automate away the junior work, and you lose the pipeline that produces tomorrow's experts. AI is not replacing them.
Check out the full recap here.
We'll be posting more in-depth takeaways on our company LinkedIn over the coming weeks.
Join us in South San Francisco for DataDrivenPharma West 2026, now 2 days, Oct 15-16.
Get your #ddpwest26 pass (presale pricing through June 30): https://luma.com/ddp_west_2026
The 5 points above are just scratching the surface, and being there in person won't be matched by any recap we put out. Presale is live now at the lowest price we'll offer, period. After June 30, prices increase.
And for the DDP regulars: any past #ddpwest25 attendee who registers by June 30 also gets a free Pharma/Biotech guest code. Bring the colleague who argues with you in meetings; they'll fit right in.
Guest codes must be redeemed within 2 weeks of receipt.
