An analysis of over half a million women in the United States confirms that artificial intelligence, applied to tomosynthesis (3D mammography), improves cancer detection without increasing false positives. The results reinforce a key premise: to make screening more effective, it is not enough to “see more,” but to see better. This is the logic behind Health Triage’s operational approach, which focuses on the reliable identification of negative cases.
In November 2025, Nature Health published the results of the ASSURE study, one of the largest ever conducted in the United States on the use of artificial intelligence in breast cancer screening using digital tomosynthesis (3D mammography). The study involved over 579,000 women and showed a 21.6% increase in cancer detection rates compared to traditional methods (without AI), with no increase in recalls or false positives.
AI, integrated into the clinical workflow as a support to the individual reader, improved sensitivity and accuracy, particularly in patients with dense breasts.
These data confirm the clinical potential of AI solutions in diagnostic imaging. However, while the ASSURE study demonstrates that AI can enhance diagnostic quality in complex cases—using tomosynthesis as a screening modality and AI as decision support for the human reader—Health Triage takes a complementary approach: safely identifying negative cases, reducing the number of exams requiring double reading and freeing up time, attention, and clinical resources for those few cases where cancer is present.
What this means for cancer prevention
The results of the ASSURE study open up new perspectives for preventive medicine:
- They demonstrate the effectiveness of AI integrated into tomosynthesis, even on a large scale and in routine clinical practice, as well as in academic studies;
- They enable the adoption of more effective screening models, especially in patients with dense breasts.
At the same time, the Health Triage model focuses on another strategic lever: reducing the workload for radiologists through the automatic identification of negative cases, which represent more than 99% of the screened population.
These different approaches are not in conflict, but complementary: on the one hand, AI is used to improve the identification of the most difficult tumors to detect, and on the other, AI is used to safely lighten the load on the system, allowing clinical staff to focus on high-priority cases.
Health Triage’s point of view
- Automatically classify negative mammograms and those to be double-read;
- Safely and verifiably reduce the workload in negative cases;
- Maintain negative predictive values close to 100%, minimizing false negatives.
It is a radical innovation in its simplicity: not a system that imposes new protocols, but intelligent technology that integrates into existing workflows and makes them more efficient, sustainable, and accurate.
