New Stool Test Detects 90% of Colon Cancers: A Breakthrough in Early Detection

Researchers at the University of Geneva have developed a non-invasive stool-based screening test for colorectal cancer that utilizes machine learning to identify the disease with 90% accuracy.

The new method analyzes the gut microbiome—the community of bacteria residing in the digestive tract—to detect early markers of cancer. According to the research, this detection rate is close to the 94% detection rate achieved by colonoscopies and outperforms all currently available non-invasive detection methods.

Precision Mapping of Gut Bacteria

The primary innovation of this screening tool is the level of precision used to analyze the microbiota. Rather than examining broad bacterial species, the research team mapped the bacteria at the subspecies level.

This intermediate resolution is designed to be specific enough to capture differences relevant to the disease while remaining consistent across diverse populations. By focusing on these subspecies, the machine-learning model can more accurately identify the presence of colorectal cancer in simple stool samples.

Comparison with Existing Methods

Colorectal cancer screening has traditionally relied on colonoscopies, which are invasive but highly effective. The University of Geneva’s stool test offers a low-cost, non-invasive alternative that maintains a high level of reliability.

The 90% detection rate represents a significant step toward the reliability of a colonoscopy. Researchers suggest that the accuracy of the model could be further increased by incorporating additional clinical data, potentially bringing it even closer to the 94% benchmark set by colonoscopies.

Clinical Implications and Future Potential

Because the test is non-invasive and low-cost, it could lower the barriers to regular screening for colorectal cancer. The use of an inventory of bacteria present in stool samples allows for a simpler diagnostic process compared to traditional endoscopic procedures.

The integration of machine learning allows the test to process complex microbial data to find the specific signatures associated with malignancy. This approach shifts the focus from general bacterial presence to the high-resolution mapping of subspecies to ensure the test remains effective across different patient groups.