World’s First AI-Native Cancer Diagnostic Test Produces Personalized Predictions with Record-Breaking Accuracy

By LabMedica International staff writers
Posted on 29 Nov 2024

Image: In a validation study, Ataraxis Breast reduced prediction error by 50% compared to standard of care genomic assays (Photo courtesy of Ataraxis AI)

Molecular diagnostic tests have long been the gold standard for selecting personalized treatments, particularly in oncology. However, these tests require physical tissue samples and come with notable drawbacks, including limited accuracy, lengthy development processes, narrow application scope, and high costs. To address these challenges, scientists are working on the development of multi-modal artificial intelligence (AI) models and diagnostic tools that can predict patients' cancer risks and outcomes, thus enabling more personalized and effective treatment options.

Ataraxis AI (New York, NY, USA) has introduced its first clinical diagnostic test, Ataraxis Breast, powered by Kestrel, the company's foundational AI model for digital pathology. This model surpasses current pathology models by uncovering novel features that are linked to patient outcomes, which are often too complex for human experts, including physicians, to fully understand. These findings span various types of diseases, and Ataraxis uses them to develop multi-modal diagnostic tests that improve in accuracy over time, can be applied swiftly to any clinical scenario, have a broad scope, and are delivered through a software-based platform. The Ataraxis Breast test leverages multi-modal patient data, including pathology slides from routine biopsy and surgery specimens, to predict patient outcomes and assist in personalizing treatment decisions across all breast cancer subtypes.

Ataraxis Breast is the first clinically validated AI-powered prognostic and predictive test for invasive breast cancer. It was developed and validated using data from 8,161 breast cancer patients treated at 15 institutions across seven countries. This study included patients with various subtypes of early-stage and locally advanced invasive breast cancer, making it one of the most comprehensive evaluations of a prognostic/predictive test. In validation against three external cohorts from leading international cancer centers, the Ataraxis Breast test reduced prediction errors by roughly 50% compared to standard genomic assays for hormone receptor-positive invasive breast cancer. Ataraxis plans to expand its diagnostic tests to cover at least 50% of the 26 million new cancer diagnoses expected globally by 2030. By introducing this new category of tests, Ataraxis is transforming cancer care, providing doctors with powerful tools to personalize treatment plans, ensuring that every patient receives the most effective care tailored to their needs.

Related Links:
Ataraxis AI