AI-Powered Blood Test Accurately Detects Ovarian Cancer

Ovarian cancer ranks as the fifth leading cause of cancer-related deaths in women, largely due to late-stage diagnoses. Although over 90% of women exhibit symptoms in Stage I, only 20% are diagnosed in the early stages (Stage I or II) because the symptoms, such as bloating, abdominal pain, and digestive issues, often resemble those of benign conditions. Current diagnostic methods, which typically rely on invasive procedures or less reliable biomarkers, frequently fail to detect the disease at an early stage. This challenge is further exacerbated by the shortage of gynecologic oncologists, which limits timely access to specialized care. Consequently, a reliable early detection test for symptomatic women is urgently needed. Now, a groundbreaking artificial intelligence (AI)-powered multi-omic platform has demonstrated high accuracy in detecting ovarian cancer in symptomatic women, a group where early diagnosis is critical but often delayed.

The GlycoLocate platform developed by AOA Dx (Denver, CO, USA) integrates multi-omic data by combining lipid, ganglioside, and protein biomarkers from a small blood sample using liquid chromatography mass spectrometry (LC-MS) and immunoassays. Machine learning algorithms analyze these complex multi-omic datasets to identify disease-specific signatures, providing results that surpass those from models relying on single biomarker types. This approach positions the test as a promising tool for clinical diagnostics. In a pioneering study, AOA Dx’s platform demonstrated high diagnostic accuracy for ovarian cancer detection, outperforming traditional markers such as CA125. In collaboration with world-renowned institutions, researchers at AOA Dx analyzed around 1,000 patient samples representing a real-world clinical population, showing strong performance in this crucial group.

Image: The GlycoLocate platform uses multi-omics and advanced computational biology algorithms to diagnose early-stage cancers (Photo courtesy of AOA Dx)

The study was conducted in two independent cohorts, both of which represented clinically similar populations. Cohort 1 was used for model training, while Cohort 2 served as an independent testing set, consisting of prospectively collected symptomatic samples from AOA’s intended use population. In Cohort 1, the model achieved an area under the curve (AUC) of 93% when distinguishing all stages of ovarian cancer from controls and 92% for early-stage (stage I/II) disease. In Cohort 2, the model maintained excellent performance with an AUC of 92% for ovarian cancer overall and 89% for early-stage disease. These results underscore the reliability of AOA Dx’s machine learning algorithms in identifying cancer-specific biomarker patterns. Additionally, previous research at AOA Dx has shown the potential clinical value of lipidomics for early ovarian cancer detection.

“Our platform detects ovarian cancer at early stages and with greater accuracy than current tools,” said Oriana Papin-Zoghbi CEO and Co-Founder of AOA Dx. “These findings show its potential to aid clinicians in making faster, more informed decisions for women who need clarity during a challenging diagnostic process.”

Related Links:
AOA Dx