Live Science · September 12, 2026

AI trained to 'think' like human pathologists may be better at spotting cancer

MedicineBiology

THE AI ANGLE

Mimicking human pathologist visual search behaviors to detect cancer in tissue slides

Researchers developed a training approach called Pathology-CoT that teaches AI to mimic how human pathologists inspect tissue samples by panning broadly at low resolution before zooming in on suspicious regions. When applied to evaluate lymph node slides for metastatic cancer, the resulting model achieved up to 100% sensitivity in identifying positive cases, outperforming general vision-language models. This shift from training on static diagnostic labels to modeling active diagnostic workflows highlights a promising pathway for developing AI screening assistants in pathology.

THE TEACHING ANGLE

Faculty can examine the clinical trade-offs between high sensitivity and elevated false-positive rates to discuss whether AI should act as an autonomous diagnostic agent or an assistive second reader that directs human attention.

Read the original at livescience.com Generate teaching or study materials

Instructors get discussion guides, assignments, and mini-cases. Students and readers get a plain summary, class prep, and an exercise. All built from the full article. Three are free with an account.

AI trained to 'think' like human pathologists may be better at spotting cancer

More in Medicine