A pathomics-based artificial intelligence model demonstrated the ability to predict which patients with metastatic non-small cell lung cancer will respond to immunotherapy. The tool, named Path-IO, was validated using data from international real-world patient cohorts and a Phase III randomized clinical trial, according to research presented at the AACR 2026 conference.

The AI model analyzes digitized pathology slides to stratify immunotherapy outcomes. Its validation across diverse, multi-national cohorts suggests the findings are robust and not limited to a single patient population or healthcare system. The research indicates the model can identify patients likely to benefit from these treatments.

While the presentation highlights the tool's predictive power, the specific regulatory pathway for such a diagnostic aid remains undefined. Further steps would likely involve seeking clearance from agencies like the FDA or EMA as a companion diagnostic to guide treatment decisions in clinical practice.

The development represents a significant intersection of computational pathology and oncology. If widely adopted, such a tool could help optimize treatment selection, potentially improving patient outcomes and reducing unnecessary exposure to ineffective therapies and their associated costs.

Oncologists have long sought reliable biomarkers to predict immunotherapy response, as current methods are imperfect. A validated AI model could address a critical unmet need in personalized cancer care, though its integration into routine clinical workflows would require demonstration of practical utility beyond retrospective validation.