What did the FDA authorize for Clairity Breast, and what did NCCN add?

AI breast cancer risk prediction reached the clinic when the FDA granted De Novo authorization in June 2025 to Clairity Breast, software that estimates five-year breast cancer risk from a routine screening mammogram. In 2026 the NCCN added an AI-derived five-year risk of at least 1.7% as one criterion for identifying higher-risk patients.

Traditional risk tools rely on age, family history, genetic testing and breast density, yet most women diagnosed with breast cancer have no known mutation or strong family history, as an American Cancer Society expert noted in the announcement. An image-based score tries to read risk from patterns already present in the mammogram.

This page explains how the tool works, what was validated, how the NCCN guidance uses it, what is not yet known, and what women considering screening or treatment in India, China or elsewhere should ask.

Key facts on Clairity Breast and AI mammogram risk scores

As announced by the company and reported by trade media:

  • FDA: De Novo authorization announced in June 2025; the authorized device name is Allix5
  • Function: predicts a woman’s five-year breast cancer risk from a routine screening mammogram, without needing family history or other clinical inputs
  • Eligibility: people without known breast cancer who are having a screening mammogram
  • Validation: trained on millions of images and validated on about 77,000 mammograms from five geographically distinct screening centres, with five-year outcome data (company-reported)
  • Equipment note: developed using Hologic 2D screening mammography systems
  • NCCN 2026: Guidelines for breast cancer screening and diagnosis include an AI mammogram-based five-year risk of at least 1.7% as one higher-risk criterion, and extend risk assessment to begin at age 35
  • First clinical use: reported at Beth Israel Deaconess Medical Center in Massachusetts

Key takeaways: AI breast cancer risk prediction and the NCCN update

  • Clairity Breast received FDA De Novo authorization in June 2025 to predict five-year breast cancer risk from a screening mammogram.
  • The 2026 NCCN Guidelines include an AI mammogram-based five-year risk of at least 1.7% as a higher-risk criterion, with recommended actions such as supplemental imaging and risk-reduction strategies.
  • The score is not a diagnosis and does not detect a current cancer; it estimates the probability of a diagnosis within five years.
  • Most public validation details, including about 77,000 mammograms, come from the company and its announcements.
  • A risk score should be discussed with a clinician who can combine it with family history, genetics, breast density and personal preference.

How does AI breast cancer risk prediction from a mammogram work?

Clairity Breast analyses pixel-level features of a routine bilateral screening mammogram that correspond with future breast cancer risk. It outputs a five-year risk score and a percentage representing the probability of a breast cancer diagnosis within five years, which is sent to the clinician through existing communication channels. The company says the score is based on the image alone, which makes it possible to deliver risk information at a normal screening visit.

Earlier research models such as Mirai, a deep learning model that estimates risk across several future time points, were evaluated in multiple international populations. In a 2021 study cited by OncoDaily, Mirai identified 41.5% of women who developed breast cancer as high risk at five years, compared with 22.9% using the Tyrer-Cuzick model.

What did the FDA authorization of Clairity Breast cover?

The FDA granted De Novo authorization, a route for novel devices with no existing comparison device, to a platform designed to predict five-year breast cancer risk from a routine screening mammogram. The authorized device name is Allix5 and the commercial name is Clairity Breast. The authorization covers use in people without known breast cancer who are having a screening mammogram.

The company says the model was trained on millions of images and validated on about 77,000 mammograms from five screening centres, including hospital-based and free-standing facilities serving diverse populations, with validation anchored on five-year outcomes. A MedTech Dive report noted that the tool was developed using Hologic 2D screening mammography systems.

Key NCCN 2026 changes for AI mammogram-based risk

As reported by OncLive in April 2026:

  • An AI mammogram-based five-year risk of 1.7% or higher was added as one criterion for higher-risk classification
  • Recommended actions after identifying higher-risk individuals include supplemental imaging and risk-reduction strategies
  • Periodic risk assessment over time is advocated
  • The population that can be identified as at increased risk was expanded to begin at age 35
  • Source: NCCN Guidelines for breast cancer screening and diagnosis, version 1.2026

What does a five-year AI risk score of 1.7% mean for screening?

A five-year risk of 1.7% means that, among women with a similar score, about 1.7 in 100 would be expected to receive a breast cancer diagnosis within five years. NCCN uses 1.7% as one threshold for identifying higher-risk patients. Reaching the threshold does not mean cancer is present; it means the woman may be offered additional steps, such as supplemental imaging and a discussion of risk-reduction options, according to the guideline summary.

The threshold is one criterion among several. Family history, genetic mutations and breast density remain part of risk assessment, and the NCCN update advocates periodic reassessment over time.

How does the AI risk score compare with family history and breast density?

Family history, genetic testing and breast density are the traditional tools, but a large share of women who develop breast cancer have no known mutation or strong family history, and density is too broad to stratify individuals, according to the American Cancer Society’s Robert Smith in the announcement. A study presented at the RSNA meeting reported that Clairity Breast gave more precise risk stratification than breast density assessment.

These are statements from an announcement and a conference abstract, not a head-to-head randomised comparison of outcomes. Clinical models such as Tyrer-Cuzick combine many inputs and remain widely used. The sensible view is that the AI score adds information about the breast tissue to existing tools, and the combination, not any single method, should guide decisions.

Does knowing your AI breast cancer risk improve outcomes?

That is not yet shown. The validation described so far measures how well the score predicts diagnoses within five years. It does not show that women who receive the score and then have extra imaging or preventive treatment develop fewer advanced cancers or die less often. Outcome studies take years because breast cancer develops slowly.

The potential benefits are earlier detection through supplemental imaging in higher-risk women, and conversations about chemoprevention or lifestyle. Potential harms include anxiety, extra imaging with false positives and biopsies, and overtreatment. Because the score is delivered during an otherwise normal screening, it also raises questions about consent, communication and equity of access.

Limits of the AI risk prediction evidence

The regulatory and guideline milestones are real, but the supporting evidence has gaps.

  • Most public validation details come from the company’s announcements, and we did not find outcome studies of acting on the score
  • The tool was developed using Hologic 2D screening mammography systems, so performance with other equipment or imaging types needs confirmation
  • It applies only to people without known breast cancer who are having a screening mammogram
  • A risk percentage is not a diagnosis, and uncertainty remains at the individual level

Who could ask about AI breast cancer risk assessment?

Access is limited and depends on the imaging centre. These are situations to raise with a clinician, not recommendations:

  • You are having a screening mammogram and have no known breast cancer
  • You have dense breasts and want to understand your risk beyond density
  • You have a family history of breast or ovarian cancer, or a known gene change
  • You are in your late 30s or 40s and unsure when to start screening or whether to add MRI
  • You were told your risk is average but remain worried
  • You want a second opinion on a risk assessment or screening plan

How do women receive an AI breast cancer risk score?

The score is calculated from a screening mammogram taken within the previous months and delivered to a clinician. One consumer health platform describes analysing a mammogram from the last 12 months and releasing provider-reviewed results with guideline-based information. Beth Israel Deaconess Medical Center in Massachusetts began offering the score to patients at one location, with plans, according to the developer, to introduce it at other clinics in Massachusetts and other states.

Availability is still limited and varies by centre and region. Ask your imaging centre whether it offers an AI risk score, who explains the result and what follow-up steps are used for a higher score.

How does Clairity Breast differ from AI mammography screening tools?

AI mammography screening tools, such as those tested in the MASAI trial, read a mammogram to detect cancer that is present now, helping radiologists decide who to recall. Clairity Breast is different: it estimates the likelihood of a future diagnosis even when the current mammogram is normal. One is detection; the other is prediction.

The two can complement each other. A woman could have a normal AI-assisted screen and still receive a higher five-year risk score, prompting a discussion about additional imaging. Conversely, a low-risk score does not remove the need for regular screening or for prompt assessment of symptoms.

AI breast cancer risk tool availability by region

Verified status and gaps:

  • United States: FDA De Novo authorization (June 2025); clinical use started at Beth Israel Deaconess Medical Center in Massachusetts, with expansion planned by the developer
  • NCCN: 2026 guidelines include an AI mammogram-based five-year risk of at least 1.7%; NCCN guidance is widely referenced, but local screening policy differs by country
  • Other research models: Mirai was evaluated in several international populations (2021 study)
  • India and China: we have not verified regulatory status or availability of Clairity Breast or equivalent AI risk scores in either country

Is AI breast cancer risk prediction available in India or China?

We have not verified whether Clairity Breast or any equivalent mammogram-based AI risk score is available, approved or used in clinical practice in India or China. The FDA authorization applies to the United States, and the model was developed using specific mammography equipment, so local availability and validity should not be assumed.

Women in India or China can still take practical steps: ask your breast imaging centre about breast density reporting, family-history risk assessment and genetic counselling, and whether supplemental MRI or ultrasound is advised for your risk.

What happens next for AI breast cancer risk prediction?

Evidence to watch includes prospective studies of women who receive the score, outcomes of supplemental imaging in newly identified higher-risk groups, performance across equipment types and populations, and cost-effectiveness. The developer says it plans expansion across health systems, and other research models, such as Mirai, have been evaluated internationally.

Risk-based approaches are also being tested in other cancers, including AI lung cancer screening on CT, and the AI-assisted reading of mammograms and AI-driven pathology tools show AI entering the breast pathway at several other points. Each has its own evidence base and its own limits.

Questions to ask your doctor about an AI breast cancer risk score

  • Does this imaging centre offer an AI risk score, and is my mammogram eligible?
  • What does my five-year risk mean in plain terms, and how does it combine with my family history and breast density?
  • Does my score change when I should start screening or how often I should have it?
  • Would supplemental imaging, genetic counselling or risk-reduction medicines be appropriate for me?
  • Who will follow up if my score is higher, and how will I be told?
  • Can my mammograms and reports be reviewed by a specialist for a second opinion?

Sources

Where the figures in this guide come from:

OncLive: Clairity Breast Is Added to NCCN Guidelines for Breast Cancer Screening and Diagnosis (April 14, 2026)

Business Wire: Clairity Becomes the First FDA-Authorized AI Platform for Breast Cancer Prediction (June 2, 2025)

MedTech Dive: Clairity receives FDA OK for breast cancer risk prediction tool

DocWire News: FDA Authorizes First AI Platform for Breast Cancer Risk Prediction

News-Medical: Clairity Breast receives FDA authorization for AI-based breast cancer risk prediction (June 3, 2025)

Cancer Discovery: AI Breast Cancer Prediction Tool Hits the Clinic (2026)

OncoDaily: How the Clairity model could change breast cancer risk prediction

GlobeNewswire: Clairity announces update to NCCN guidelines to include AI mammogram-based risk tool (April 14, 2026)

Frequently asked questions

Is Clairity Breast FDA approved?

Clairity Breast received FDA De Novo authorization in June 2025, under the device name Allix5, for predicting five-year breast cancer risk from a routine screening mammogram in people without known breast cancer. Strictly, this is a device authorization rather than a drug approval, and it applies to the United States.

What does an AI breast cancer risk score tell me?

It estimates the probability of being diagnosed with breast cancer within five years, based on patterns in your screening mammogram. It is not a diagnosis and does not find a current cancer. A higher score may prompt a discussion about supplemental imaging or risk-reduction options with your clinician, alongside family history, genetics and breast density.

What is the 1.7% AI risk threshold in NCCN guidelines?

The 2026 NCCN Guidelines for breast cancer screening and diagnosis list an AI mammogram-based five-year risk of at least 1.7% as one criterion for identifying higher-risk patients. Recommended actions include supplemental imaging and risk-reduction strategies. It is one criterion among several, and NCCN also advocates periodic reassessment of risk over time.

Does an AI risk score replace family history or genetic testing?

No. Family history, genetic testing and breast density remain part of risk assessment. The AI score adds information from the image itself, and a conference study reported more precise stratification than density alone. No sources reviewed show that the AI score replaces other tools, so decisions should combine all available information.

Is AI breast cancer risk prediction available in India or China?

We have not verified whether Clairity Breast or an equivalent AI risk score is available or approved in India or China. If you live there, ask your imaging centre about density reporting, family-history risk assessment and genetic counselling.

What should I do if my AI risk score is high?

Ask your clinician what the score means for you and what follow-up is recommended, such as supplemental imaging, genetic counselling or risk-reduction options. A high score is not a diagnosis, and most women with a higher risk will not develop cancer within five years. Consider a specialist second opinion if the plan is unclear.

How is Clairity Breast different from AI that reads mammograms for cancer?

AI mammography screening tools, such as those tested in the MASAI trial, read a mammogram to detect cancer that is present now. Clairity Breast is different: it estimates the likelihood of a future diagnosis, even when the current mammogram is normal. One is detection; the other is prediction.

The two can complement each other. A woman could have a normal AI-assisted screen and still receive a higher five-year risk score. A low-risk score does not remove the need for regular screening or for prompt assessment of symptoms.

Does knowing my AI breast cancer risk score improve outcomes?

That is not yet shown. Validation so far measures how well the score predicts diagnoses within five years. It does not show that women who get the score and then have extra imaging or preventive treatment develop fewer advanced cancers or die less often.

Possible benefits include earlier detection through supplemental imaging in higher-risk women. Possible harms include anxiety, extra imaging with false positives and biopsies, and overtreatment. Discuss the score with a clinician who can combine it with your family history, genetics and breast density.

This page is general information, not medical advice. OncoVOIS explains and helps you prepare; your oncologist decides about your care.