Breast cancer: This Nvidia-powered AI tool predicts cancer recurrence risk before surgery — here's how it works

A new artificial intelligence Nvidia-powered tool could help doctors assess the risk of breast cancer returning years before a recurrence occurs. SimBioSys, a precision medicine company, has received US Food and Drug Administration (FDA) clearance for TumorSight Risk, software that uses breast MRI scans, pathology and clinical data to estimate an individual patient's risk of cancer recurrence over five and 10 years.

The company describes TumorSight Risk as the world's first AI-powered, MRI-derived prognostic tool for assessing breast cancer recurrence risk. Designed to work with information already collected during routine care, the software aims to give oncologists additional insights to support treatment planning, potentially before surgery.

The AI tool is intended for adult women with hormone receptor-positive (HR-positive), human epidermal growth factor receptor 2-negative (HER2-negative) Stage I to IIIA breast cancer. According to SimBioSys, this subtype accounts for approximately 75% of breast cancers in the US.

How does the new AI breast cancer tool work?

TumorSight Risk uses AI models to analyse information from breast MRI scans alongside pathology and clinical data. Rather than relying exclusively on conventional tumour measurements or tissue-based tests, the software examines tumour characteristics and the surrounding tissue context to generate an individualised estimate of recurrence risk.

The cloud-based platform produces a score ranging from zero to 100. It also assigns patients to one of three categories: low, intermediate or high risk.

The score is designed to estimate recurrence-free survival at five and 10 years, providing doctors with an additional source of prognostic information when discussing treatment options with patients.

The company says the tool can provide prognostic information before surgery, potentially allowing doctors to incorporate recurrence-risk estimates earlier in the treatment-planning process.