Clairity
Clairity Breast estimates a woman's probability of being diagnosed with breast cancer within five years of a bilateral screening mammogram. The prescription software reads compatible directly acquired 2D images from Hologic Lorad Selenia and Selenia Dimensions systems, looking for patterns in the tissue that a reader cannot see, and returns a percentage after the radiologist has read the screening study. The risk estimate sits apart from the radiologist's assessment of the current exam and gives no care recommendation. The FDA granted De Novo authorization for the device, as Allix5, in May 2025, and 2026 NCCN guidelines added AI based mammogram risk assessment as an option from age 35.
In the FDA review's retrospective study of 77,511 exams at five US sites, five year area under the curve was 0.70 overall and 0.66 after excluding cancers found within 90 days, with calibration limits in some groups, including women with implants. Beth Israel Deaconess Medical Center scored its first clinical patient in February 2026, Invision Sally Jobe offers it in Colorado, and since October 1, 2026 Everlywell has sold a $249 assessment nationwide in which Clairity retrieves eligible images after the patient authorizes release and a clinician reviews the result. Health system installations receive mammograms from the imaging archive and return the score through HL7.
Radiologist Connie Lehman, a Harvard Medical School professor, founded the Boston company in 2020 and became chief executive in January 2026. ACE Global Equity and Santé Ventures led its $43 million Series B in November 2025, and Clairity Breast 3D and Clairity Heart are in development.
Capability Axes
An AI Health Index grade measures what a buyer can verify from public sources on the date shown. It is not a rating of how good the product is. A vendor can build an excellent system and grade low on an axis because it publishes nothing an outsider can check. How grades read
The device uses a computer vision model to generate a five year risk estimate from a previously acquired screening mammogram. The model examines the four standard views from a compatible exam and returns a probability. Clinical sites receive that number in their existing system after a radiologist reads the images. The prediction is the function the practice buys. Other services around it include image retrieval and clinician review in the Everlywell channel. The assessment itself depends on the model's analysis of features in the screening images and gives clinicians a separate result they can discuss with the patient.
The FDA indication places the risk score after radiologist interpretation of the screening mammogram. The software sends the result automatically into a clinical information system and gives no care recommendation. Clinical action remains with the provider. In the Everlywell channel, a board certified clinician reviews the score before release to the patient's account. The FDA review also describes limits on supported images and warns that some higher estimates can overstate risk. The provider workflow does not give an error or uncertainty cutoff for suspending delivery of the result, or a single documented override and escalation procedure across both channels.
The FDA review describes a computer vision deep learning algorithm trained on screening mammograms from 22 clinical sites, hospitals and freestanding facilities among them. Biopsy confirmation served as the positive outcome reference, with later imaging or the absence of biopsy confirmed disease used for negatives. The device registry lists Clairity Breast version 10.4.5 in commercial distribution. That gives a version identifier and describes where the training data came from. The model's architecture is not specified, and no release history shows how changes reach imaging sites as the company develops a 3D version for tomosynthesis.
The authorization describes the Allix5 image model and ties its indication to a specific prescription device, compatible images and use after radiologist interpretation, and the FDA device registry names an exact model version. A cleared model of that kind cannot be quietly swapped.
The consumer route adds Everlywell and the imaging facility, with Clairity obtaining images once the patient authorizes release; those parties supply records or care and are distinct from the model that scores the exam. No subprocessor list identifies every recipient of patient images, the hosting provider or the terms for storage and any downstream processing.
The FDA review provides a retrospective validation set of 77,511 screening exams from five sites, with biopsy confirmation for positive cases. The area under the curve for discrimination at five years was 0.70 overall and 0.66 when exams associated with a cancer identified within 90 days were excluded. The second analysis narrows the contribution of cancers present close to screening. This was an evaluation of prediction against later diagnoses in an existing dataset. It did not assign patients to a new screening pathway or measure outcomes after clinicians changed care in response to the score.
In clinical installations the system receives mammography images from an imaging archive, and in the Everlywell route the patient signs a medical records release so Clairity can retrieve images from the facility, with a clinician reviewing the result before it appears in the patient's account. Those steps explain where an exam comes from.
No retention period is given for submitted images or scores, no deidentification method is described, and nothing commits the company not to use submitted images for training future models such as Clairity Breast 3D. The FDA review sets input and performance limits, but there is no stated process for handling a safety event or an incorrect risk score.
The consumer channel uses a records release so the screening facility can send images to Clairity, and in health systems the product runs under each organization's own intake and privacy practices. Those processes establish how a particular exam is requested. No service privacy document covers how the company handles images for the risk assessment, no business associate role is stated for any deployment, and no agreement is offered.
The company's site links to a privacy policy and a safety page without a HIPAA position. A hospital's or reseller's own privacy notice describes that organization's duties and does not set the manufacturer's obligations for processing a mammogram.
The device authorization describes software safeguards for the intended imaging workflow and the performance testing behind them, and consumer intake requires the patient's permission before an exam is requested. Neither amounts to an examination of the manufacturer's own security operations. There is no trust portal, SOC 2 report, ISO certification, HITRUST attestation or outside security testing schedule for the image processor.
The support material says who retrieves an image and who reviews the result. It gives no scope for an independent assessment of the environment that stores and processes those mammograms, which reach Clairity from health systems and, through Everlywell, from facilities nationwide.
The FDA granted De Novo DEN240047 on May 30, 2025, for Allix5, a Class II prescription device under 21 CFR 892.8500. The device registry lists Clairity Breast version 10.4.5 in commercial distribution. The authorized use is a five year breast cancer risk estimate from compatible directly acquired 2D bilateral screening images, reviewed after the radiologist has interpreted the exam.
FDA's labeling excludes synthetic 2D images and states that the output does not detect or diagnose a present cancer or give care recommendations. The indication applies to patients without known breast cancer at screening.
The FDA review compares discrimination and calibration by race and ethnicity, age, breast density, implant status and equipment. Five year area under the curve was 0.67 in Black participants and 0.73 in White participants, with separate estimates for Asian and Hispanic patients.
The review also identifies groups where calibration fell outside the study's prespecified bounds, and in patients with breast implants the uncertainty in estimated performance was too large to meet the acceptance criterion. Those results describe variation within the retrospective validation. They do not show whether screening decisions made on the score, such as added imaging, would perform similarly across those groups.
A radiologist interprets the original mammogram separately, and a clinician can review the risk result alongside other information. The Everlywell channel adds provider review before release and a refund when a mammogram cannot be retrieved or scored. That policy addresses failed delivery of an assessment, not a completed estimate later found to be wrong.
There is no route by which a patient can ask the company to recalculate a score, no accuracy warranty for the result, and no commitment to compensate someone whose care followed a mistaken estimate. The FDA labeling's statement that some higher estimates can overstate risk is a disclosed limit rather than a remedy.
The authorized workflow receives mammograms as DICOM files from the site's imaging archive and sends the score back over HL7. The patient does not need a separate interface for clinician review at a participating facility. Beth Israel Deaconess confirms the product is offered in its breast imaging service. Hologic Lorad Selenia and Selenia Dimensions are named compatible image sources.
That information does not identify a named electronic health record with verified reads and writes at the hospital. The consumer channel returns an assessment through the Everlywell portal after clinical review rather than through the imaging system.
The hospital product joins a clinical imaging workflow by accepting a mammogram from an image archive and sending the result back to clinical systems, so providers order, review and document the score inside their existing imaging and reporting tools. Everlywell's intake lets Clairity request an image from the facility and sends the reviewed result to a patient account. Neither route says where the manufacturer runs the image processing engine.
No cloud region, on site hosting option, data residency commitment or tenant isolation is described for the device. The source imaging facility and the destination record are clear, while the infrastructure that processes the mammogram between them is not.
Everlywell lists $249 for one Clairity Breast assessment, opens intake from the product page and has sold it nationwide since October 1, 2026. The retail fee covers the partner's pathway, including image retrieval and a clinician reviewed result, and Everlywell refunds an order when the image cannot be obtained or analyzed. Medical centers contract with Clairity to use the software in their screening workflows, and the Series B funds reimbursement work aimed at community imaging centers.
The institutional fee, any per assessment charge to a health system and the cost of configuring imaging and reporting interfaces are not given. The public retail price sets a unit for the patient channel only.
The device is intended for breast screening workflows that produce directly acquired 2D mammograms on compatible Hologic systems, and its FDA validation includes hospital based and freestanding screening sites. Beth Israel Deaconess Medical Center offers it through its breast health service and scored its first clinical patient in February 2026, Invision Sally Jobe offers it in Colorado, and Everlywell provides consumer access from an existing eligible mammogram.
The indication is a five year risk estimate after screening in patients without known breast cancer at that exam, and 2026 NCCN guidelines name AI mammogram risk assessment as an option from age 35. The validation does not cover synthetic 2D images or establish performance for patients with breast implants.
Alternatives to Clairity
Other products that do the same kind of work as Clairity, and what each one does.
ScreenPoint Medical, founded in 2014 by Nico Karssemeijer and Sir Michael Brady and based in Nijmegen in the Netherlands, makes Transpara, breast AI for mammography and tomosynthesis.
Genius AI Detection is the breast imaging AI software line of Hologic, Inc. of Marlborough, Massachusetts, licensed to breast imaging buyers separately from Hologic's wider diagnostics, surgical and skeletal businesses.
Lunit is a cancer focused medical AI company founded in 2013, headquartered in Seoul and listed on KOSDAQ. Its radiology line is Lunit INSIGHT: MMG for 2D mammography (FDA cleared 2021), DBT for tomosynthesis (cleared 2023) and CXR for chest X-ray (CXR Triage cleared 2021).
Better Medicine develops software that flags, measures and characterizes renal lesions on CT scans. BMVision Kidney uses deep learning to locate cystic and solid lesions and generate findings a radiologist can inspect in BMViewer.
Enlitic builds software that cleans up and standardizes medical imaging data for hospitals, radiology groups and archive operators.
Company founded by interventional radiologists to find peripheral arterial disease in imaging already taken. Its one product in market, VascularAssist Occlusion Triage (VAOT), was cleared through FDA 510(k) K260729 in June 2026 as Class II triage and notification software.
Pricing
Vendor-published figures are labeled as such. Figures labeled “Estimated” are derived from third-party sources and have not been confirmed by the vendor.
| Entry Price | Pricing Basis | BAA Tier | Implementation | Source |
|---|---|---|---|---|
|
$249 per assessment through Everlywell
$249 baseline
|
Per consumer assessment through Everlywell, nationwide since October 1, 2026. Health systems contract directly, with institutional pricing set in each agreement. | No business associate role or agreement is stated for any deployment. | Institutional interface and setup fees are not listed. | Vendor Published |
Everlywell lists a single Clairity Breast assessment at $249, covering image retrieval and a clinician reviewed result, and refunds an order when the study cannot be retrieved or scored. Clinical site licensing fees are not listed.