AI-Powered Precision for Risk of Recurrence Prediction

Ataraxis Breast provides personalized recurrence risk prediction for stage I–III breast cancer patients. By analyzing morphological features from standard H&E slides together with clinical data, the test delivers both a precise risk estimate to help personalize care.

5-year recurrence risk across all subtypes:

Deeper prognostic insight for personalized treatment planning

Introducing next-generation technology for prediction of breast cancer recurrence

No more ambiguity

By providing a clear high or low 5-year recurrence risk results, Ataraxis Breast helps resolve uncertainty in challenging and inconclusive cases, such as patients considered to be intermediate risk or borderline by standard clinical characteristics and genomic tests.

For all molecular subtypes

Ataraxis Breast's recurrence risk prediction was developed and validated using data from all early-stage breast cancer patients, including HR+/HER2-, but also triple-negative (TNBC), HER2-positive, and triple-positive breast cancer.

A different perspective

Ataraxis Breast's recurrence risk prediction was found to provide independently prognostic information, unique from standard clinicopathological features and genomic assays.

How do we predict risk of recurrence?

Flowchart showing Ataraxis AI Models using H&E slide and clinical data to predict 6% low risk and 31% high risk of 5-year recurrence with corresponding treatment recommendations.

Flowchart showing use of H&E slide and clinical data to assess Ataraxis risk: 6% low risk with recommendation to consider treatment de-escalation, and 31% high risk with recommendation to consider treatment escalation.

Data you already have

Our platform uses the existing H&E-stained biopsy or resection slide collected during diagnosis, along with standard clinical variables such as:

Tumor size and nodal status
ER, PR, and HER2 status
Age and histologic subtype

Multimodal AI Analysis

The model evaluates thousands of micro-regions across the pathology slide—capturing tumor morphology, spatial architecture, stromal features, and immune contexture—and integrates these imaging-derived features with clinical data using the Ataraxis multimodal AI architecture.
‍
This enables a high-resolution, individualized estimate of 5-year recurrence risk across HR+, HER2+, and triple-negative disease.

Predictive Output

The Ataraxis Breast RISK report provides:

5-year recurrence probability (%)
A validated Low vs High risk grouping (no intermediate zone)

These outputs offer a clear, actionable view of each patient’s recurrence risk to guide treatment and follow-up decisions.

Development & Clinical Validation

Robust development and validation data

Ataraxis Breast’s recurrence risk prediction capability was developed and validated using data from 8,161breast cancer patients treated at 15 institutions in 7 countries.  

The study included patients with all subtypes of early-stage and locally advanced invasive breast cancer, marking this as one of the most comprehensive evaluations of a prognostic/predictive test.

Kaplan Meier curves comparing low and high risk patients

Comparison of Ataraxis Breast to 21 gene assay

Over 50% improvement in accuracy compared to standard of care genomic tests

In validation across three external cohorts from leading international cancer centers, Ataraxis Breast reduced error in predicting 5-year recurrence risk by approximately 50% compared with standard-of-care genomic assays for hormone receptor-positive invasive breast cancer.

Works in all clinically relevant invasive breast cancer subgroups

Ataraxis Breast’s recurrence risk prediction capability was developed and validated on robust and diverse data from 7 countries. The data included previously underserved groups, such as triple-negative breast cancer patients. Ataraxis Breast was shown to be consistently prognostic across groups with various age, race, molecular and histological subtype, and nodal status.

Ataraxis Breast stratifies patients in challenging subgroups where genomic assays fail

Premenopausal node-positive (N=218)

Invasive Lobular Cancer (N=627)

Two calibration plots showing Kaplan-Meier estimates of observed 5-year recurrence probability versus median predicted probability, comparing uncalibrated and spline models against perfect alignment; top for all patients and bottom for patients with legacy genomic testing.

Superior calibration to real-world outcomes

Traditional prognostic tools, including widely used genomic assays, often provide relative risk scores. These scores indicate who is higher or lower risk within a population, but they do not reliably match the true recurrence probabilities patients experience in practice.

Ataraxis Breast's 5-year recurrence risk prediction is fundamentally different. It is designed not only to rank patients by risk, but to produce absolute risk estimates that reflect what actually happens in real-world cohorts.

What this means in practical terms:

Older approaches often don’t reflect real-world rates. Published validations of legacy assays show poor alignment between reported risk and actual outcomes; a “15% risk” may behave very differently in practice. Our test addresses this mismatch.

The predicted Ataraxis test score is trustworthy. If Ataraxis Breast predicts a 12% recurrence probability, the real-world recurrence rate in similar patients is very close to 12%. Our scores don’t over- or under-estimate the risk.

Consistency across all breast cancer subtypes. Whether the tumor is HR+, HER2+, or TNBC, Ataraxis Breastmaintains strong calibration, ensuring that a predicted percentage has a consistent clinical meaning across disease types.

Why calibration matters clinically: A well-calibrated model allows oncologists to act on the absolute value of the prediction, not just its relative position, making treatment escalation or de-escalation safer, clearer, and more evidence-based.

View a Sample Report

Explore our sample report and learn more about how predictions made by Ataraxis Breast can help in recurrence risk assessment and making more personalized decisions about treatment.

See Report

Make the most informed call for your patients.

What is my risk of cancer coming back in 5 years?

Get a clearer recurrence risk result without the wait or the gray zone

Legacy recurrence tests, such as the majority of genomic tests, can take weeks to get a result and still leave many patients in an unclear “intermediate” category. Ataraxis Breast is designed to address those exact issues, and it can be used for early-stage breast cancer across:

All molecular subtypes, regardless of ER, PR, or HER2 status
Premenopausal women and postmenopausal women
Regardless of cancer presence in lymph nodes

We start with the tissue your care team already collected during biopsy or surgery. Our advanced AI platform analyzes thousands of microscopic tumor features, along with basic clinical information, to estimate your personal risk of breast cancer recurrence.

Female doctor showing information on a tablet to a woman patient in a consultation.

I might want this test if...

I have triple-negative (TNBC) or HER2+ breast cancer

I want a better sense of whether my risk of recurrence is lower or higher, and I want a prognostic tool validated across breast cancer subtypes.

My genomic test result is unclear (e.g., "intermediate")

I want additional insight to help decide on treatment—especially if I'm premenopausal or have lymph node involvement.

I am anxious and want to see quicker results

I want to reduce extra weeks of anxiety and get a clear result sooner, so I can start my treatment plan as soon as possible with my care team.

I want the most accurate risk estimate possible

I want the most realistic recurrence risk estimate for my specific cancer type and feel confident that I am not being over-treated or under-treated.