Dr. Caroline Chung, recognized by Reuters as a Trailblazing Woman of 2026 in Enterprise AI, served as the inaugural Vice President and Chief Data & Analytics Officer at The University of Texas MD Anderson Cancer Center, co-directing the Institute for Data Science in Oncology and holding a tenured professorship in Radiation Oncology. A clinician specializing in CNS malignancies, she translates clinical challenges into enterprise-wide AI strategy, bridging precision medicine and patient outcomes at scale. Her global leadership spans co-president of the Quantitative Medical Imaging Coalition, co-chair of ASCO’s AI Community of Practice, advisory roles with the NIH and NCI, and co-authoring NASEM’s Digital Twins report. Dr. Chung is a defining voice in responsible, impactful AI implementation in medicine.
Recently, in an exclusive interview with CIO Magazine, Caroline shared insights into how clinical experience, quantitative rigor, and culture-first leadership are converging to redefine cancer care. On AI trends in oncology for 2026, she flagged multimodal AI as truly transformational while calling for more research on AI implementation, impact measurement, and the human-AI interface to understand how learning and critical thinking evolve. As Chair of Women in Cancer – All in Cancer, she credited “Strengthening Through Perspectives” events for moving the needle by fostering open dialogue across clinicians, researchers, STEM, industry, and patient advocates, stressing that sponsorship opens doors, because women and underrepresented leaders are often over-mentored and under-sponsored. The following excerpts are taken from the interview.
Hi Caroline. Twenty years in radiation oncology and quantitative imaging is a foundation few AI leaders share. Can you take us back to the moment you first saw the link between pixels, data, and a patient’s life, and how did that shape your path?
It was not a single moment but an accumulation of lived experiences in the clinic, challenges identified in research and the promise of emerging technologies to leverage data to enable a better future for our patients.
As a radiation oncologist, defining the target for radiation treatment is a core part of delivering effective treatment. Very early in my career, I started in pursuit of extracting much more information from imaging data than just anatomy, information about the underlying biology. Information that could tell us which areas of the tumor would be most likely to respond to treatment or be most likely to recur and may benefit from treatment intensification. However, treating a pixel as more than just an image signal but rather a measurement demands a metrology that supports quantitative medical imaging.
While we generate enormous amounts of imaging data across medicine, we currently continue to treat a lot of this data as visual aids to clinical decision making when it could be used as quantitative measurements. With the growing capabilities of AI and computational algorithms, there is great opportunity to utilize quantitative imaging measurements to support precision medicine, increasing the speed of novel therapeutic discovery and the efficiency of clinical trials to advance clinical care. Precision medicine relies on precision measurements and this applies to current clinical care, even more with the integration of AI into clinical care and critically to enable digital twins.
What do you love the most about your current role?
What an extraordinary opportunity it has been to take on the role of the inaugural Chief Data Officer and then Chief Data & Analytics Officer, which meant a unique opportunity to start with a blank slate and ask: what should this actually look like to maximize the impact to serving our mission to end cancer?
Focusing on our people first and bringing mutual conversations around the processes to ensure that technology enables and supports the end goals has been a core approach. Even when it comes to AI, I’ve written about how ‘Culture, not code, is the core of every AI strategy” (https://www.forbes.com/councils/forbestechcouncil/2025/12/01/culture-not-code-is-the-core-of-every-ai-strategy/).
Working beyond the technical questions such as which algorithm to deploy, but rather how do we work collectively to build an organization where people can readily find, appropriately access and trust data, where they feel capable and equipped to ask better questions of it, where they see themselves as active participants in a learning health system to iteratively improve? That’s the work that keeps me energized. Because a truly future-ready health system isn’t built on dashboards or models alone, it’s built on a shared belief, across clinicians, researchers, administrators, and patients, that the data we generate and how we generate it can and should make care better each and every day. And recognizing that a patient journey is often not isolated in a single institution, it will take much broader collaborations across systems and stakeholders to realize the full potential.
AI in oncology moved from hype to clinical trials fast. From your seat, what is the single trend in AI for cancer care that’s truly transformational in 2026, versus still experimental?
If I had to name one trend that has great promise to bring practice-changing impact, it’s multimodal AI. Although there is further work to be done in this arena, meaningful integration of pathology, radiology, genomics, and clinical data into a single analytical layer is showing great potential and meaningful integration across these data domains has incredible powerful for informing clinical decisions and revealing new insights.
There are so many areas that need further research, development and experimentation. One that I will call out is the need for more research around the implementation of AI in healthcare and the measurement of impact, which was also recently highlighted in Nature. Many publications to-date have focused on model performance and accuracy, but this is only part of what is needed to drive to impact. An additional area that needs much more research is the human-AI interface. Better understanding how this dynamic relationship evolves our learning, critical thinking, perceptions and evolution of both knowledge and beliefs can help us mitigate risks and maximize impact.
Quantitative imaging and predictive modeling are converging. How are these tools changing the standard of care for early detection, and where are health systems still underprepared?
For most of modern medicine and oncology until today, the clinical imaging workflow has allowed for the generation of heterogeneous imaging data as long as a radiologist is able to visually read the images and provide a report on what they see, including an evaluation of whether the tumor(s) are progressing or regressing. With this approach, medicine currently remains reactive.
As we start to leverage predictive modelling to move from reactive medicine to anticipatory clinical decision making, what we’re witnessing is a fundamental repositioning of imaging in cancer care from a qualitative, adjunctive data source to a quantitative measurement platform. That distinction sounds technical, but the clinical and operational implications are enormous.
Take early detection as an example. AI-based models using digital breast tomosynthesis have shown the promise to forecast five-year breast cancer risk from routine screening images. There are AI models being utilized to anticipate the need for cardiac catheterization. A number of these models are in clinical use already today.
But here’s where health systems are still deeply underprepared: the lack of standardization. Quantitative imaging practices remain inconsistent across institutions, scanners, and protocols. You can’t build a predictive model on top of data that isn’t reproducible. Until we treat image acquisition with the same rigor we apply to laboratory assays we will keep leaving signal on the table.
Beyond this fundamental and pragmatic aspect is the gap between detecting a signal and connecting it to the underlying biology. The opportunity to use quantitative imaging to create in vivo biological signal from non-invasive imaging is incredible. We have barely scratched the surface of what imaging can tell us about the tumour microenvironment, molecular phenotype, and treatment response without a single biopsy. The major challenge with radiomics or radiogenomics has been finding signal that can persist amidst all the technical noise that is generated across non-calibrated imaging studies.
Diversity in leadership is central to your mission. As Chair of Women in Cancer – All in Cancer, what practices have actually moved the needle on mentorship and inclusive pathways?
Several years ago, we started an event series under the theme “Strengthening Through Perspectives” and we have continued it past the initial year because that statement is so central to our mission of supporting the growth of leadership through mentorship, sponsorship, networking and learning resources across the entire community supporting cancer – clinicians, researchers, educators, administrators, those in STEM, regulatory and industry as well as patient advocacy. The diversity of perspectives comes from the different educational backgrounds, work and personal lived experiences, areas of expertise that help us look at situations and the world differently. When open, constructive dialogue around a shared goal occurs across a table of individuals with these different perspectives, mutual learning, respect and deeper understanding of each other and the shared challenges occurs thereby leading to inspiration, synergy and momentum. Beyond the group benefit, when individuals at the table realize they are not alone, and they also stop interpreting structural barriers as personal failures. That cognitive shift is underestimated as a force multiplier.
In order to build this broad perspective and set the table, we have encouraged both mentorship and sponsorship to bring differing perspectives to leadership teams and strategic discussion. Mentorship gives advice. Sponsorship opens doors. Research on this has found that women and underrepresented leaders are commonly over-mentored and under-sponsored. We’ve worked hard to encourage that leaders go beyond guiding by advocating, nominating, and pulling people into rooms they wouldn’t otherwise enter.
Art, music, or nature often inform how we think about systems. What’s a personal favorite that captures your view of “precision” — in life or in medicine — and why?
Fractals are genuinely universal. You find them in the branching of a river delta and the branching of a bronchial tree, in the spiral of a nautilus and the spiral of a galaxy, in the recursive complexity of a fern and the recursive rhythms of a Bach fugue. The universe, it turns out, repeats its own logic and vibration across every scale of existence. What is a better reflection of precision.
In oncology, we have spent a great deal of time categorizing histologies, staging and outcome responses. But what quantitative imaging and data science are revealing is that the boundaries are not so clear and that by embracing the complexity, we may identify paths that would have been blurred out through oversimplification. For instance, a tumor margin isn’t a clean edge and a patient’s response isn’t necessarily binary. Precision isn’t about simplification, it’s about finding the right level of resolution at which the truth reveals itself.
This perspective also applies to leadership. Culture isn’t just a few key words that everyone can recite, it’s built in patterns that repeat at every scale, from how a leader talks about failure in a meeting, to how a frontline analyst feels when they raise an uncomfortable question. If the pattern is not supported down to the smallest scale, the culture does not hold at the large scale.
Fractals remind me that complexity and elegance are not opposites. The most precise thing you can do is find the pattern that was always there at the scale that is fit-for-purpose.
Mentors shape more than careers. Who is someone — in or outside oncology — you consider a role model for inclusive leadership, and what trait of theirs do you emulate?
I have been fortunate to have a number of incredible mentors in my life both in oncology and outside of oncology and even outside of medicine. A trait that I have seen consistently across my mentors is genuine curiosity and asking good questions to learn and listen. Emulating this quality has helped me be mindful and present in the moment, make deeper connections, open up incredible worlds and opportunities that were unanticipated and enriched my life and journey as a life long learner.
Aspiring leaders see your title and want the roadmap. For a young professional at the intersection of medicine, data, and tech, what three skills should they master first, and what myth should they ignore?
I get this question a lot, and I always resist providing a list of skills or capabilities that would show up on a job description as it was certainly not a planned straight path to where I have come. Instead, I will share some deeper core skills that I feel are valuable and that I continue to work on each day.
The first is translational fluency, the ability to move between worlds without losing your footing in either and connecting and compounding strengths across disparate domains. A logical example is the ability to speak oncology in one room and data architecture in the next while ensuring all involved in the conversation can relate to the content such that they are all solving the same problem. Another example is listening to a talk about the swarming nature of insects and applying the concepts and observations to team culture building around data science.
The second is effective communication. Invest time to learn, practice and continue to grow your ability to communicate clearly across different audiences for different purposes. Finding your authentic voice, across all media of communication, to help share your vision and gain understanding will help you along whatever journey you take. The ability to drive to impact and scale impact relies on people to support successful execution and implementation. For this, it is important to remember that communication is a two-way street and proactive and effective ways of sensing and listening are just as key. Every successful project, program, or organization that I have helped build started with people who believed in the mission and vision and collectively took the actions needed to make it real.
The third is the ability to ask critical questions and discern promising signal from lots of noise. The pace of technology, particularly in data and AI, is moving incredibly quickly and while there is much excitement and promise, there is also an incredible amount of hype and exaggerated success. Being able to dig a level deeper to ask critical questions from the perspective of the data, the technology and the clinical application will help you discern. To this point, ignore the idea that depth and breadth are in tension, that becoming a domain expert means you’ll lose your edge across disciplines. The most powerful positions in this space, the ones shaping how AI actually transforms cancer care or health systems, are held by people who went deep enough to be taken seriously and broad enough to be trusted to lead. You don’t have to choose. The intersection is a key position of strength.
What is your biggest goal? Where do you see yourself in 5 years from now?
It is helpful to have a Big Hairy Audacious Goal (BHAG), one that generates a sense of urgency and focus. Mine is to positively impact the lives of at least one billion people. When your north star is that scale, you stop asking ‘what’s the next logical step?’ and start asking ‘what’s the highest-leverage thing I can do to move towards that goal? How can I best utilize my skills, experience and expertise, network and time to intentionally make progress towards this goal?’ The everyday decisions such as what to build, what to publish, where to invest my energy look completely different when you hold them against that horizon.
Health data, AI, and computational medicine are converging at a moment when the decisions we make in the next few years will shape care delivery for the next few decades. I look forward to contributing at tables where these decisions are made, both advising and architecting. Whether that’s through a learning health system that reaches underserved communities, through AI frameworks that become policy, or through platforms that put precision medicine within reach far beyond academic medical centers, the common thread is scale with equity.
The billion isn’t a vanity number. It’s a compass. It keeps me honest about whether what I’m collectively growing and building with others and how I’m investing my time is truly transformative.
Looking at the arc from radiation oncology to leading enterprise AI, if you had to capture your professional journey in one sentence, what would it be, and what chapter are you writing next?
“My career has been grounded by a belief that the biggest impact isn’t made alone nor in a single step, it’s made by connecting knowledge, resources and people through shared mission and collective learning.”
I started by building cross-functional teams, programs and solutions within Radiation Oncology as I pursued the introduction of MR in radiotherapy and then helping build out one of the first multidisciplinary brain metastases clinics and programs in the world in Toronto. Then expanded beyond any disease site domain or department to institutional.