The AI Revolution in Breast Cancer: Predicting Oncotype DX Results from Pathology Slides
In this article
A breakthrough overcoming cost and logistical barriers: analysing routine pathology slides with deep learning makes it possible to determine the need for chemotherapy with 90% accuracy.
Dynamic architecture: the “hidden” genomic signatures on the slide
Oncotype DX produces a risk score (Recurrence Score - RS) by examining the expression of 21 genes. The AI model is based on the principle that a tumour’s genomic structure is actually reflected in its morphology. Microscopic details missed by the human eye are clear sources of genomic data for deep learning algorithms.
B) TAILORx Clinical Trial Design: Shows patient classification by recurrence score and randomisation of intermediate-risk patients.
D) Relationship Between AI Scores and Oncotype DX: Documents the high correlation between AI scores and actual recurrence scores (Pearson r = 0.728).
The model was built on a foundation model trained on more than 171 thousand histopathology slides. This enabled AI to learn the “language of pathology” perfectly.
The model was trained using data from TAILORx, one of oncology’s most reliable randomised clinical trials. This proved that AI did not merely make a “prediction”, but forecast “clinical benefit”.
Clinical findings: as precise as the genomic test
The study showed that, in identifying patients classified as high risk (RS ≥ 26) by the Oncotype DX test, artificial intelligence achieved an AUC of 0.898 , an outstanding performance.
H, I) Sensitivity and Specificity Analyses: Curves showing the model’s clinical performance at different binarisation thresholds.
A major shift in risk classification
Among patients considered “high risk” according to traditional clinical criteria (tumour size, stage, etc.; MINDACT criteria), especially postmenopausal women, 31.3% were reclassified by AI as “low risk” . Follow-up data confirmed that these patients derived no benefit from chemotherapy. This result means that tens of thousands of women can be “precisely” spared the severe side effects and financial burden of chemotherapy.
| Performance Criterion | Artificial Intelligence (AI) Result | Clinical Meaning |
|---|---|---|
| High-Risk Prediction (AUC) | 0.898 | 90% concordance with the genomic test. |
| Low-Risk Negative Predictive Value | 0.982 | When AI says “low risk”, the test is also low in 98% of cases. |
| Chemotherapy Benefit (Premenopausal) | HR 0.63 | Chemotherapy is lifesaving in the group AI identifies as high risk. |
Why is this a turning point? Speed, cost and access
From a health informatics perspective, the technological leap offered by this model rests on three core pillars:
In contrast to a genomic test costing 3500 dollars, digital scanning costs less than 1 dollar .
Instead of a 2-3 week logistics and waiting period, results appear within minutes on the screen.
Of patients beyond the reach of genomic tests, 95% can now regard “precision oncology” as more than a dream.
Particularly in India, Africa and other developing regions, this technology will bring a genuine wave of democratisation to breast cancer management.
DROZDOGAN Academy Commentary
For years, we sought healing in breast cancer by “decoding the cell’s genetic code in external laboratories”. This study, however, shows us that this code is actually present like a “picture” on the slide in front of the pathologist. In keeping with our philosophy of biological restoration, the real issue is being able to read the clues left by nature using the most precise and accessible technology.
Here, AI is not a “decision-maker” but an extraordinary “guide” that strengthens the physician’s compassion and scientific precision. Genomic medicine is about to cease being the exclusive privilege of wealthy countries and enter the service of all humanity. Healing will be built at that precise crossroads where the intelligence of science meets the democratising power of technology.
Scientific References and Further Reading
- 1. Shamai G, et al. Deep learning on histopathological images to predict breast cancer recurrence risk and chemotherapy benefit. The Lancet Oncology. March 11, 2026. doi:10.1016/S1470-2045(25)00727-2
- 2. Sparano JA, et al. Adjuvant chemotherapy guided by a 21-gene expression assay in breast cancer (TAILORx). N Engl J Med. 2018.
- 3. Kalinsky K, et al. 21-gene assay to inform chemotherapy benefit in node-positive breast cancer (RxPONDER). N Engl J Med. 2021.
Was this article helpful?
Your feedback helps us improve our content.
Reader comments
Share your comment
Your name and comment are published only after administrator approval. Your email address is never shown to visitors.
Be the first to comment.