Throne Science presented its first clinical validation study at the 2026 American College of Gastroenterology (ACG) Annual Scientific Meeting, evaluating an AI model that classifies stool using the Bristol Stool Scale (BSS).
The study compared the model’s classifications with assessments from five gastroenterologists using 278 real-world images collected from beta testers across all seven BSS categories. For the most common stool type, BSS Type 4, the AI model achieved 89.6% exact-match agreement, compared with a 59.9% human-to-human agreement baseline.
Throne’s toilet-mounted sensor is designed to passively track bowel movements, hydration and urinary function, providing continuous data that could help clinicians understand patients’ symptoms between visits. The company aims to reduce reliance on patients’ recall and self-reporting when tracking digestive health.
The findings represent an initial validation, not evidence that the technology improves diagnosis or patient outcomes. Throne plans larger real-world studies to evaluate its performance further. Its longer-term goal is to develop at-home screening for early signs of colorectal and urinary tract cancers, but the reported study did not assess cancer detection.