Reimagining Oncology with Artificial Intelligence
Artificial intelligence (AI) is no longer a futuristic concept confined to science fiction. In the world of oncology, it is becoming an indispensable tool, helping clinicians navigate the ever-expanding universe of medical knowledge. Brian P. Mulherin, MD, medical director of the American Oncology Network, emphasized this transformation during the inaugural MiBA Community Summit. He noted that AI is increasingly essential to manage the overwhelming influx of new data and treatment guidelines in cancer care.
“There’s an increasing need to help somebody, at scale, organize all this and deliver these insights to the clinician at the point of care,” Mulherin stated, highlighting the challenge faced by community oncologists who are often inundated with information yet lack support systems to process it effectively.
The Exponential Growth of Oncology Knowledge
Medical knowledge is growing at an astonishing rate, with new information emerging approximately every 73 days. The approval of new drugs, such as imlunestrant for estrogen receptor-positive, HER2-negative, ESR1-mutated advanced breast cancer, adds to the complexity. Mulherin questioned how many oncologists would feel confident prescribing such treatments only days after FDA approval, underscoring the pressing need for timely and actionable insights.
With over 30,000 ongoing clinical trials in oncology globally and around 50 new drug approvals annually, the necessity for AI-driven clinical decision support tools is more evident than ever.
AI and Large Language Models in Oncology
Mulherin explored the role of large language models (LLMs) like ChatGPT and Llama 2 in summarizing and interpreting oncology guidelines. He cited a 2025 study assessing the performance of these models in answering complex clinical questions based on the National Comprehensive Cancer Network (NCCN) guidelines for thyroid carcinoma.
The results were mixed. While both models showed potential, their responses ranged from fully correct to partially or completely incorrect, suggesting that current LLMs, though promising, are not yet reliable enough for high-stakes medical decision-making. Other LLMs such as Gemini, Claude, and Perplexity also face similar limitations, primarily due to their generalist training and lack of access to proprietary medical data like NCCN guidelines.
“These models are trained on the entire corpus of human knowledge,” Mulherin explained, “but they still lack the specificity and access to critical sources that would make them truly effective in clinical oncology.”
Domain-Specific AI Tools and Their Limitations
Turning to more domain-specific tools, Mulherin discussed OpenEvidence, a platform that aggregates medical data from journals, abstracts, and conference proceedings. While useful, it too falls short by not incorporating NCCN guidelines, revealing a significant gap in AI utility for oncology.
This limitation highlights the importance of developing AI tools that not only access wide-ranging data but also integrate authoritative clinical resources to enhance decision-making.
AI in Biomarker Testing and Genomic Profiling
One area where AI has shown tangible benefits is in the identification and monitoring of biomarker testing. The NCCN recommends next-generation sequencing (NGS) for over 14 biomarkers in patients with advanced non-small cell lung cancer (NSCLC). AI can analyze institutional data to determine who is being tested and ensure no patient falls through the cracks.
At Hematology Oncology of Indiana, where Mulherin practices, implementing AI-driven notifications for comprehensive genomic profiling significantly improved testing rates. Before the program, 87.34% of stage III and IV NSCLC patients had undergone NGS. Seven months post-implementation, that number jumped to 99.41%.
“AI can prompt physicians: ‘Your patient has an EGFR mutation but isn’t on osimertinib—why not?’” Mulherin said, likening AI to a digital safety net that ensures best practices are followed.
The Future of AI in Oncology
Despite its limitations, AI is already playing a pivotal role in enhancing the quality and equity of cancer care. Mulherin believes that future iterations of AI will not replace physicians but will act as indispensable partners in delivering personalized, evidence-based treatment.
“Even with substantial improvements in the technology itself, AI will work in conjunction with physicians to improve the quality of care, ensure equitable access, and provide data for future research and development,” he concluded.
While AI in medicine is still evolving, its impact in oncology is both significant and growing. From simplifying complex guidelines to boosting biomarker testing compliance and improving clinical decision-making, AI is poised to become a permanent fixture in the oncologist’s toolkit.
This article is inspired by content from Original Source. It has been rephrased for originality. Images are credited to the original source.