On Wednesday, February 15, 2023, ACCC President Dr. David R. Penberthy hosted his final Tech Talk: “The Impact of Big Data and Artificial Intelligence on Oncology.” More than 60 individuals—a balanced mix of ACCC cancer program and industry members—registered for this event, with about 42 attending live. This interactive discussion featured four panelists:
Each panelist shared their perspectives on how technologies like artificial intelligence (AI), machine learning, and the use of big data are currently impacting cancer care delivery, as well as their vision of the ideal future state. To start, Dr. Baehner spoke on Exact Sciences’ molecular diagnostic test(s) and the role big data plays in cancer diagnostics.
As a practicing pathologist, Dr. Baehner addressed the need to validate upcoming cancer diagnostic assays like next-generation sequencing to ensure its usability and efficacy in clinical practice. “Selecting the genes is one component of developing of new diagnostic tests,” he said. “The other thing…is making sure that the assay is analytically validated.” With the creation of Exact Science’s Oncotype DX Breast Recurrence Score®—a diagnostic test used to detect 16 breast cancer-related genes in an individual and inform treatment options—Dr. Baehner and his team completed several clinical validation trials over a 10-year period before implementing the test into everyday practice. Since, this assay has been used for more than 1.5 million people to inform treatment decisions for better patient outcomes.
Dr. Penberthy then turned the conversation over to Dr. Frownfelter, who discussed newer and timely technologies that have great potential to positively disrupt healthcare delivery as we know it today—from remote patient management (more than just patient monitoring) to the digital human and Microsoft Chat GPT. “Animating a human avatar in a way that is interactive with us and that makes it [the avatar] feel like a human-to-human interaction is pretty remarkable,” he said. “What would I do with that? Why would I care about that as an oncology practice?”
According to Dr. Frownfelter, the digital human is innovating the way interactions between humans and technology feels, such as interacting with a chat bot online, a bot over the phone, or even another person. Further, Dr. Frownfelter believes this technology can improve patients’ health literacy. “If you combine this [the digital human] and…Chat GPT and generative AI…. then you’ve got an avatar that’s interacting in a very human-like way with the data built into it to inform how it responds, based on the use case or conversations,” he said, adding that “the case for health literacy is pretty strong.” This technology can help reduce medication errors, hospitalizations, and, ultimately, mortality because people may be in a better position to receive optimal care when they present to the emergency department or hospital.
Moving the conversation into machine learning, Dr. McGough shared more on how this technology can impact the many stages of drug development, specifically in precision medicine. She explained that investment in data and advanced analytics should be the priority for transforming healthcare at all levels of drug development. In using databases like the Flatiron-Foundation Medicine Clinico-Genomic Database, companies like Genentech are given the bigger picture of the real-world patient—their clinical history and tumor genomic profile—which, in turn, informs the development of targeted therapies. “We no longer have to study cancer in silos because these databases contain dozens of different cancer types and tens of thousands of patients,” Dr. McGough said. “This concept of tumor agnostic or pan tumor research is really just changing the paradigm of how we understand cancer.”
This is just one example of the first steps the research community is making toward advancing precision medicine for patients with cancer—understanding and treating the disease on the molecular level. Machine learning then comes into play to help identify the most important predictors of survival across various cancer types and patient populations. “We can train machine learning models to predict survival using thousands of clinical and genomic variables that we can obtain from patient health records,” Dr. McGough said. “We can create heatmaps to show you what variables are showing up strongly prognostic, either with a harmful effect or a protective effect on survival.” This type of technology can then be used to risk stratify patient populations and help inform treatment decisions in the clinic.
In wrapping up the day’s discussion, Dr. Adamson featured a picture of her team at Flatiron Health. “This is the team of cross functional machine learning engineers at Flatiron that are working with me every day,” she said. “Together we are building these large language models that are able to read things in a similar way.” She affirmed that there will never be a replacement for the providers who must read patients’ charts and understand how test results impact treatment decisions. But technology is being created to assist these providers in identifying the biomarker much quicker.
“Just imagine a semi-truck full of paper documents where there’s just been all this wealth of information trapped within…the technology developments for the last couple of years are really what is now going to open up the opportunity for us to learn,” Dr. Adamson said. To do so, the ideal future state will take teams of engineers, healthcare professionals, researchers, and more, working together to innovate and fully implement technology to its fullest potential to truly transform the way cancer care is delivered today.
You can access the on-demand recording of this Tech Talk on the ACCC website.
We welcome you to share our blog content. We want to connect people with the information they need. We just ask that you link back to the original post and refrain from editing the text. Any questions? Email Maddelynne Parker, Content Manager.
To receive a weekly digest of ACCCBuzz blog posts each Friday, please sign up in the box to the left.