Dr. Vasan Yegnasubramanian, MD, PhD, serves as the director of Precision inHealth Medicine at Johns Hopkins. In a recent discussion with the communications team from the Johns Hopkins Kimmel Cancer Center, he explored the transformative impact of artificial intelligence (AI) on cancer research and treatment. Below is a summary of their conversation.
Why is AI crucial in cancer research and treatment?
Artificial intelligence (AI) is set to revolutionize health care significantly. Rather than just adding another tool to our medical arsenal, it challenges us to rethink our entire approach to medicine. Although data is an invaluable resource, its true power is realized only when it is refined and utilized effectively. AI facilitates predictive, real-time, and consistent healthcare delivery, paving the way for innovations previously thought impossible. From early cancer detection to personalized treatments and expedited drug discovery, AI holds transformative potential, provided we approach its application responsibly and ethically.
Can you elaborate on the four primary priorities?
I often liken data to oil: both are incredibly valuable but only when processed appropriately. The first priority is ingestion, which involves identifying and extracting high-value data from sources like medical records or imaging databases. While data in records aids clinical care, it must be harnessed for innovation.
The second priority is annotation and curation, which entails structuring, cleaning, and organizing the data to make it research-ready. This step is akin to refining oil, converting raw material into a vital fuel for innovation.
Priority three focuses on establishing distribution channels and computing platforms. Instead of sending large datasets to researchers, which poses risks and inefficiencies, we bring researchers and their tools to the data, ensuring both privacy and efficient performance.
Lastly, priority four emphasizes responsible stewardship. Unrestricted access to sensitive data is not an option. As we wield this power, it is crucial to implement governance to prevent harm. At Johns Hopkins, we have created an AI and Data Trust Council for oversight.
How are you addressing privacy concerns and fostering public trust?
We are committed to maintaining robust security. Security risks escalate when data is decentralized and stored on individual computers, as a malicious actor needs to breach only one machine. Therefore, we are transitioning to centralized, monitored systems where access is logged, and any unusual activity is promptly flagged. We also have a Chief Information Security Officer who is instrumental in identifying and preventing breaches and implementing immediate response systems. Our objective is to maximize data availability for research while ensuring patient anonymity and protection.
What does public trust mean in this context?
Much of our data was originally collected for clinical purposes. Now, we are repurposing it for secondary research, which involves both legal frameworks and an ethical commitment to demonstrate that our actions genuinely benefit society. Engaging the public as a partner is vital. This requires transparency, thoughtfulness, and inclusivity—building trust is essential for progress.
What possibilities does this approach unlock for science?
The exciting part lies in interdisciplinary collaboration across medicine, data science, and engineering, leading to full-scale transformation in health care. Here are some key advancements:
- Shifting from reactive to predictive care, enabling early disease detection and timely intervention.
- Transforming erratic care into real-time care, where AI can continuously monitor data from wearable devices, flagging issues and alerting care teams instantly.
- Minimizing unwanted variations, ensuring consistent application of best practices across different healthcare settings.
Moreover, we are accelerating innovation. AI can uncover patterns in imaging that are beyond human detection, screen billions of virtual drug compounds, and even design new treatments from scratch—all while significantly shortening development timelines. This is just the beginning.
How do these innovations benefit individual patients?
What we are developing caters to both scale and personalization. While the datasets we utilize are extensive and anonymized, the insights derived contribute directly to personalized care. This reflects the essence of a learning healthcare system—collective data informs individual care, creating a cyclical model where innovation drives improved patient outcomes.
What about the risk of overdiagnosis or overtreatment with AI?
This is a valid concern, especially regarding early detection. Just because we can identify issues earlier doesn’t mean treatment is always warranted. AI assists in aligning early detection with risk stratification—helping us determine whether a detected anomaly is potentially harmful. For instance, we are working on biomarkers in prostate cancer to differentiate between aggressive and non-threatening cases, ensuring patients receive the appropriate level of care. Similar strategies will be developed for every type of cancer.
How is this being integrated back into healthcare delivery?
We are creating tools like Patient Insight, a dashboard that contextualizes a patient’s journey against thousands of others. Another initiative, the Health General Reasoner, is designed to embed AI within clinical workflows. We are also piloting a collaboration with Microsoft on a tool developed at Johns Hopkins that helps identify patients at risk for blood clots, a common complication of hospitalization, and suggests preventative treatment in real time.
Are there plans to extend AI initiatives beyond the Kimmel Cancer Center?
No single institution has access to all the necessary data or expertise. Therefore, we are part of a national initiative called the Cancer AI Alliance, collaborating with four other leading cancer centers to create a federated learning platform. This model allows each center to keep its data secure while sharing AI models, thereby ensuring privacy and security while fostering extensive collaboration. If successful, it could serve as a blueprint for cancer hospitals nationwide.
Is Johns Hopkins at the forefront of this movement?
We certainly are, but it’s a collective effort. We are collaborating across institutions, with industry partners, and within our communities. While the technology is powerful, it’s the people and teamwork that will truly drive impact.
The future is bright. We are investigating how AI can hasten drug development, tailor treatments, and deepen our understanding of cancer in unprecedented ways. Simultaneously, we are committed to ensuring that this technology is inclusive, equitable, and trustworthy. If harnessed properly, AI has the potential to elevate the standard of care for everyone. This is our mission.
Explore more online:
Podcast featuring Vasan Yegnasubramanian:
https://cancer-matters.blogs.hopkinsmedicine.org/2024/12/19/cancer-matters-bill-nelson-artificial-intelligence/
Pancreatic cancer:
https://www.hopkinsmedicine.org/inhealth/pancreatic-cancer
Prostate cancer:
https://www.hopkinsmedicine.org/inhealth/prostate-cancer
