Caroline Chung, Vice President and Chief Data and Analytics Officer at MD Anderson Cancer Center, shared a post on LinkedIn:
“I’m often asked: how did a radiation oncologist end up in a data and data science role?
Excited to be featured in CIO Magazine to share some of that story and how I came to emphasize the value of building culture-first AI for cancer care. The title lands on something I’ve come to believe after a career spent between oncology, imaging science, and data: an AI strategy is only ever as strong as the culture that stewards it.
I learned this in the clinic, not the lab. Some of the most capable algorithms I’ve seen didn’t fail because the math was wrong. They failed because the teams around them didn’t trust the output, couldn’t integrate it into their workflow, or couldn’t define what needed to be measured to show its value and its risk.
Precision medicine relies on precision measurement, and precision measurement depends just as much on how we ask the question and define the measure, how we prepare the data, and whether the patient stays at the center of every design choice.
Too often AI gets treated as a destination: deploy the tool, declare victory.
The real work is quieter. It’s asking the right question before reaching for a model. It’s the discipline to tell a promising algorithm that its context of use isn’t ready yet. Sometimes that’s about the readiness of individuals (e.g. AI literacy). Sometimes it’s re-coordinating a multifunctional team so it can actually succeed. Sometimes it’s an organization maturing enough to invest intentionally – resourcing governance and mentorship across technical and clinical teams so change can take hold. That kind of culture doesn’t announce itself.
In the interview I talk about seeing fractals everywhere. These patterns repeat at every scale, from a river delta to a bronchial tree. Culture works the same way. It isn’t a few words everyone can recite; it shows up in how a leader talks about failure in a meeting, and in how a frontline analyst feels when they raise an uncomfortable question. If the pattern doesn’t hold at the smallest scale, it won’t hold at the largest. The collaboration that works at the bedside is the same collaboration that has to hold in your data governance, your model validation, your leadership.
We’re at a genuine inflection point. AI can help move us from reactive medicine to anticipatory care. But only if we build it carefully, together, and with discipline. Culture, not code.
Grateful to my colleagues at UT MD Anderson and the IDSO team, to those who learn alongside me in the AI Community of Practice at ASCO, and to the many others who’ve journeyed with me and helped shape my thinking. The biggest impact is never made alone.”
Other articles featuring Caroline Chung on OncoDaily.