New York-based medtech startup Ataraxis AI has emerged from stealth mode, announcing $4 million in seed funding.
The company has a focus on developing artificial intelligence (AI) driven cancer diagnostics that can help clinicians to pick the best therapy for their patients and predict their progress.
The company was founded in 2022 by current CEO Jan Witowski, MD, PhD, and Krzysztof Geras, PhD, who were previously both based at New York University, and has funding from Giant Ventures and Obvious Ventures.
It has developed a foundation AI model called Kestrel for use in digital pathology testing. These kinds of AI models are able to process information from multiple data sources and can be quickly adapted to different tasks. Kestrel has already been trained with millions of images from cancer tissue samples.
At the same time as announcing its launch and seed funding, Ataraxis also launched its first test, Ataraxis Breast, which the company says is the world’s first AI-native prognostic/predictive test for breast cancer.
Ataraxis Breast uses data from pathology slides using tissue taken at biopsies and other surgeries, as well as other patient information, to make predictions about patient outcomes. The test was trained and validated using data from more than 8,161 breast cancer patients from seven different countries. The cohort included people with all subtypes and stages of breast cancer.
In comparison with a commonly used genetic test called Oncotype DX, Ataraxis Breast was 30% more accurate at predicting cancer recurrence. It was also able to accurately reclass patients originally classed as “intermediate” risk into either low or high risk groups, so their treatment and follow-up could be adjusted accordingly.
“Ataraxis has created a breast cancer outcome prediction system that produces personalized predictions with record-breaking accuracy,” said company advisor Yann LeCun, PhD, Jacob T. Schwartz Professor of Computer Science at New York University, chief AI scientist at Meta, and Turing Award Laureate, in a press statement.
“They are leveraging the latest developments in deep learning and AI such as self-supervised training and joint embedding architectures. Beyond this success, Ataraxis has an ambitious vision to bring about the transformative potential of AI in healthcare.”
The company wants to develop more tests using Kestrel to cover as many cancer subtypes as possible in the future. As well as predicting outcomes for patients, they also want to incorporate specific questions into the tests. For example, if patients who have had chemoendocrine therapy can benefit from treatment with CDK4/6 inhibitors.
“Tissue biopsies provide one of the richest sources of medical data, and Jan and his team have developed a transformative approach using multi-modal AI models to help doctors identify treatment options beyond chemotherapy,” said Rohan Ganesh, partner at Obvious Ventures.
Ataraxis is by no means the only AI-based diagnostics/prognostics company in the cancer space, digital pathology is on the rise and many startups have launched during the last five years. For example, Stratipath, Auspex Diagnostics, OncoGenomX and Digistain all use AI to help predict cancer outcomes. Whether Ataraxis can come out ahead of the competition remains to be seen.