Skip to main content
5 minutes reading time (962 words)

Insights from the State of AI Precision Oncology Conference

By Waqas Haque, MD, MPH

In a virtual seminar held on December 12 of last year, the State of AI Precision Oncology brought together a group of scientists and oncologists to delve into the transformative impact of artificial intelligence (AI) in oncology. Dr. Doug Flora, the Executive Medical Director of Oncology Services at St. Elizabeth Healthcare and Chief Editor of the new journal, AI in Precision Oncology, curated discussions on early cancer detection, drug discovery, and data management.

The webinar commenced with a fireside conversation between Dr. Flora and Erik Topol, a renowned cardiologist and Founder of the Scripps Institute who is recognized as an AI thought leader. Dr. Topol emphasized the challenges that physicians face, including the encroachment of "pajama time" on patient care. AI can alleviate these challenges by improving clinical documentation, incorporating patient reminders, and coaching physicians to enhance sensitivity in patient communication. Dr. Topol stressed the importance of clinicians leading AI innovation to avoid a repeat of technological innovations that do not prioritize patient care, such as with electronic heath records.

1. Early Cancer Detection: Dr. Nikhil Thaker, a Radiation Oncologist at Capital Health, highlighted the pivotal role of AI in early cancer detection. He discussed Sybil, a validated deep learning model focused on radiomics, capable of predicting the future risk of lung cancer with remarkable accuracy. It can run in the background of a radiology reading station as lung computer tomography (CT) imaging becomes available without need for inputted clinical data or radiologist annotations. The potential for self-triage by patients in the realm of early cancer detection was underscored, emphasizing the need for physicians to spearhead innovation.

Dr. Tom Beer, the Chief Medical Officer of Exact Sciences, spoke about the company's efforts in multicancer early detection (MCED) and how this first emerged through academic and industry collaboration. DETECT-A is the first-ever large prospective interventional study to evaluate MCED through a blood test; 17 (65%) of the identified cancers were diagnosed at an early stage, with the majority of diagnoses not having a current recommended screening test for the general population.

2. Drug Discovery: Sarah McGough, a Data Scientist at Genentech, explored the intersection between algorithms and oncologists in drug discovery. She acknowledged the challenge of accounting for biases in AI models and cited the "middle of the night" testing phenomenon: patients who are sicker in the hospital will receive more detailed clinical care, which can lead the AI to find a spurious correlation between alarming lab measures (such as elevated white blood cell count) and patient survival. AI's potential in evaluating off-label targeted therapies and bridging the gap between algorithms and oncologists was a key theme.

3. Data Management and Treatment Planning: Blythe Adamson, a Principal Scientist at Flatiron Health, delved into risk prediction models in oncology—emphasizing the importance of transparency in AI models. Bethany Hills, a lawyer specializing in life sciences and health care, discussed how most AI training data comes from the three states with a heavy biotech presence: New York, Massachusetts, and California. Hills discussed how we can achieve improved transparency of AI models with saliency maps (reporting features most relevant to an outcome), gradient-based methods (showing different inputs to show how the output changes), and counterfactual explanations. Dr. Arturo Loiaza-Bonilla, the Co-Founder of Massive Bio, spoke about the role of generative AI in promoting value-based care.

Parting Thoughts: My key takeaway from the webinar is that AI is poised to revolutionize oncology. Its ultimate benefit will depend on how it is harnessed. It's not surprising to have a techno-optimistic set of speakers for a conference on AI, but caution was advised regarding the safe use of generative AI and the need for transparency in AI models.

Oncologists should learn about AI through active participation in expert-led events, engagement in projects, and following thought leaders in the field. It's important to ask about your specific interest in AI—is it improving the patient experience, improving adherence to guideline-based recommendations, or augmenting drug development?

In conclusion, the State of AI Precision Oncology webinar offered useful insights into the current landscape and future possibilities of AI in oncology. The collaboration between clinicians, data scientists, and industry leaders showcased the potential for positive transformation, with the caveat that responsible use and transparency must be at the forefront of integration in health care. The journey towards unlocking the full potential of AI in oncology has only begun, and events like these serve as milestones in the ongoing dialogue between oncology and technology.

About Dr. Haque

Waqas Haque, MD, MPH, is a third-year Internal Medicine Resident at New York University (NYU) in a Clinical Investigator Track. He recently matched to the University of Chicago for fellowship, which he will be beginning later in 2024. As a Clinical Investigator Track Resident, Dr. Haque has balanced his patient care work with a variety of research projects. He hopes to begin fellowship training next year in Medical Hematology/Oncology at an academic program with opportunities to further his work in innovative clinical trial design, value-based care delivery to cancer patients, and becoming an early-stage clinical investigator.

For More Information

Saag HS, Shah K, Jones SA, et al (2019). Pajama time: working after work in the electronic health record. J Gen Intern Med, 34(9):1695-1696. DOI:10.1007/s11606-019-05055-x

Mikhael PG, Wohlwend J, Yala A, et al (2023). Sybil: a validated deep learning model to predict future lung cancer risk from a single low-dose chest computed tomography. J Clin Oncol, 41(12). DOI:10.1200/JCO.22.01345

Lennon AM, Buchanan AH, Rego SP, et al (2023). Outcomes in participants with a false positive multi-cancer early detection (MCED) test: results from >4 years follow-up from DETECT-A, the first large, prospective, interventional MCED study. J Clin Oncol, 41(suppl_16). DOI:10.1200/JCO.2023.41.16_suppl.303

Related Posts