Artificial intelligence is moving beyond simply detecting cancer to understanding the complex biology inside tumors, revealing patterns that human pathologists cannot see with conventional microscopes. New research presented at the American Association for Cancer Research (AACR) Annual Meeting 2026 shows that AI-powered analysis of tumor tissue can identify which patients will benefit most from targeted therapies, potentially sparing others from ineffective treatments .

What Can AI See in Cancer Tissue That Doctors Cannot?

Traditional pathology relies on pathologists examining tissue slides under a microscope to assess tumor characteristics. But AI systems trained on millions of tissue images can detect spatial patterns and relationships between cells that are invisible to the human eye. In one study conducted with Agilent Technologies and Ajou University Medical Center, researchers analyzed over 25,000 non-small cell lung cancer (NSCLC) samples using AI software called Lunit SCOPE IO. The analysis revealed that tumors with high c-MET expression showed a significant reduction in immune cells within 30 micrometers of tumor cells, indicating a spatial immune exclusion pattern that conventional analysis methods missed . This finding suggests that some tumors actively create a protective barrier against the immune system, which could explain why certain patients don't respond to immunotherapy.

The implications are profound. If doctors can identify which patients have this immune-evasion pattern, they could recommend combination therapies that target both the tumor's growth signals and its immune-suppressing mechanisms, rather than relying on immunotherapy alone.

How Are AI Biomarkers Changing Treatment Selection?

One of the most compelling examples comes from an exploratory analysis of the MOUNTAINEER clinical trial, which tested a combination therapy for metastatic colorectal cancer. Researchers used AI to quantify HER2 protein expression in tumor tissue and found a striking dose-dependent relationship: patients with higher HER2 expression had response rates as high as 80%, compared to 43.4% overall . Even more striking, patients with low immune cell infiltration showed zero response to the treatment and faced significantly higher disease progression risk.

This type of AI-driven biomarker analysis could fundamentally change how oncologists make treatment decisions. Instead of prescribing the same therapy to all patients with a particular cancer type, doctors could use AI analysis to identify which patients are most likely to benefit, potentially saving time and reducing exposure to ineffective treatments with serious side effects.

Steps to Integrate AI Biomarker Analysis Into Cancer Care

• Tissue Sample Preparation: Hospitals must ensure that tumor tissue samples are properly collected, preserved, and digitized into high-resolution images that AI systems can analyze accurately and consistently.
• AI Platform Integration: Cancer centers need to integrate AI analysis software into their existing digital pathology workflows, requiring investment in infrastructure, staff training, and validation studies to ensure accuracy in their specific patient populations.
• Clinical Decision Support: Oncologists must learn to interpret AI-generated biomarker reports and incorporate these insights into multidisciplinary tumor board discussions, combining AI findings with clinical judgment and patient preferences.
• Ongoing Research Collaboration: Institutions should partner with AI developers and pharmaceutical companies to validate biomarkers in real-world settings and contribute data to larger studies that improve AI accuracy over time.

The research being presented at AACR 2026 spans multiple cancer types and therapeutic approaches. Beyond the lung cancer and colorectal cancer studies, Lunit researchers are presenting work on tumor-infiltrating lymphocyte (TIL) analysis in collaboration with Yale University School of Medicine, AI-based target discovery for bispecific antibodies, and biomarker research in CD47-targeted therapies . This breadth suggests that AI tissue analysis is becoming a foundational tool across oncology, not just a niche application.

Why Is Investment in AI Oncology Talent Critical Right Now?

Recognizing that AI and precision oncology require a new generation of researchers, UC San Diego's Moores Cancer Center has launched the inaugural Illumina/Moores Cancer Center Fellowship in Artificial Intelligence and Precision Oncology, with support from the Illumina Corporate Foundation . The fellowship is designed to accelerate research at the intersection of AI, clinical genomics, and precision oncology, with a focus on integrating complex data to improve cancer outcomes.

The inaugural fellow, Aaron Boussina, PhD, an assistant professor in the Division of Biomedical Informatics at UC San Diego, was selected for his work applying AI to improve early cancer detection. His research integrates large-scale electronic health record data, genomics, and advanced machine learning models to identify patterns that may enable earlier diagnosis of disease . Boussina brings a decade of experience from the biotechnology and biopharmaceutical sectors, where he supported genomic testing and data-driven approaches.

"Each individual has a unique health story. Multi-modal AI approaches allow us to better interpret that story and enable earlier cancer detection and intervention. This funding will help us apply these tools to drive meaningful improvements in patient care," said Boussina.

Aaron Boussina, PhD, Assistant Professor in the Division of Biomedical Informatics at UC San Diego

Boussina's work focuses on early-detection approaches, particularly for pancreatic cancer, where most cases are diagnosed at advanced stages. His AI-driven, multi-modal models integrate clinical data, imaging, and genetic information to identify cancer at earlier, more treatable stages . This represents a shift in how AI is being applied to oncology, moving beyond treatment selection to earlier intervention.

"Our AACR presentations reflect how AI is increasingly translating into real-world clinical impact. Across these studies, we demonstrate how AI-driven biomarkers can enhance precision, deepen our understanding of tumor biology and increasingly support treatment decision-making in clinical practice," said Brandon Suh, CEO of Lunit.

Brandon Suh, CEO of Lunit

The convergence of AI and genomics is creating new opportunities for cancer care, but it also requires a pipeline of trained researchers and clinicians who understand both the technology and the clinical context. Fellowships like the one at UC San Diego are critical investments in building that workforce .

As these AI tools move from research settings into clinical practice, the stakes are high. The ability to predict which patients will respond to specific treatments could spare thousands of patients from ineffective therapies while directing resources to the treatments most likely to work. The research being presented at AACR 2026 suggests that AI-powered tissue analysis is no longer a theoretical promise, but an emerging clinical reality .