You made it. Thanks for pulling up.
Scanned the QR code mid-talk? Respect. Everything from the session, the slides, the science, the speaker, and how to keep this conversation going, lives right here.
The Deck
Using Digital Twins and AI to Reimagine Cancer Treatment
- Speaker
- Robert Karienye
- Conference
- Deep Learning Indaba
- Date
- 21 Jul 2026
- Size
- .pptx · 2.3 MB
Yours to keep, remix into your notes, whatever. Just keep OncoTwin's name on it if you're passing it around. That's the whole ask.
The TL;DR
Cancer treatment protocols get validated on trial populations, then applied to one specific patient who didn't look much like the average of that trial. We talked through how mechanistic, patient-specific modelling closes that gap: how OncoTwin builds a digital twin from a patient's own clinical data, why AI's job here is narrowly to calibrate that model rather than to predict outcomes on its own, and why none of it works if it needs a stable internet connection to run.
What to Walk Away With
Robert Karienye
Founder & Research Lead at OncoTwin, based in Nairobi. Building the mechanistic models and calibration approach behind the platform, with hospitals running on limited connectivity in mind from day one.
karienye@oncotwin.africaAsk Us Anything (Kind Of)
Starting with what people actually asked us after the talk. More gets added as more of you ask.
Wait, so does AI replace the mechanistic model?
Nope. AI calibrates the mechanistic model's parameters to a specific patient and quantifies the uncertainty of that estimate. The actual model of tumour dynamics stays mechanistic and inspectable. No black box.
Does OncoTwin make the treatment call, then?
Also no. It projects likely outcomes for paths a clinician is already weighing. The clinician always makes the final decision, full stop.
The Pics
Conference photos land here once we're back. Check back soon 📸
Don't Be a Stranger
Drop your email and we'll ping you when new research lands, when the next talk is happening, or when there's genuinely something worth your inbox space.