Immunotherapy has given patients with advanced non-small cell lung cancer new hope. It helps by boosting the body’s own immune system to target cancer cells, but it only works for about 20% to 30% of patients. Most patients build up a resistance to it. Right now, doctors rely on standard biomarkers to guess who will benefit, but those tests have limits.

“We need smarter tools,” said thoracic oncologist Marina Garassino, MD, Professor of Medicine at UChicago Medicine and senior author of the study.

In a new study, researchers with the international I3LUNG project gathered clinical, imaging, pathology, and genomic data from 2,396 patients across six countries. They used all that information to build AI models that predict how well someone will respond to treatment.

Better Predictions for Lung Cancer Treatment

Photo: Butusova Elena/Shutterstock

Results showed how AI models consistently outperformed standard clinical tests. In machine learning, a score called AUC measures how accurately a model classifies information. Scores between 0.8 and 0.9 are considered excellent.

An AI model using just clinical and blood data scored 0.77. When researchers added imaging and pathology data into the mix, the score jumped to 0.88.

Advertisement

Doctors and AI Working Together

The study also examined how humans and AI work as a team. Twenty doctors, half lung cancer experts and half from other medical fields, reviewed 100 real patient cases. First, they looked at them on their own, and then they checked them with AI support.

With the AI tool, the doctors got much better at detecting who would respond to treatment. Their accuracy score went from 0.72 up to 0.87. Doctors who were not lung cancer specialists saw the biggest improvement.

“This alignment between machine and clinical logic is essential for building trust in AI-assisted decision-making,” Garassino said.

The project is now moving forward by enrolling more than 2,000 additional patients to test these tools in real-time clinic settings.

“I3LUNG establishes a new benchmark for AI in thoracic oncology. Decision support tools built even from routinely available clinical data can outperform the biomarkers we rely on today,” Garassino said. “For patients, this means fewer missed opportunities for treatment benefit. For community physicians, it means access to expert-level guidance at the point of care.”

Garassino added, “For the field, it provides a rigorous, fair, and explainable framework — validated across diverse healthcare systems and populations — that can serve as a global platform for the next generation of precision immunotherapy.”