Early-stage accuracy
96%+ stage 1–2 accuracy
An internal 10-cancer retrospective study (n=691, 5-fold cross-validation) showed 96%+ accuracy detecting stage 1 & 2 cancers.
Turn a simple urine sample into early-stage cancer insight
Toby Oncology’s AI-powered, urine-based platform is built to detect cancers at stages 1 and 2 using lab workflows that already exist.
A urine-based, multi-cancer platform
Urine chemistry + AI for early detection
Toby Oncology’s platform combines urine chemistry and machine learning to look for cancer signals earlier in the disease course. It’s designed to be:
Urine-based
Uses a standard urine sample collected at home or in-clinic, no needles or imaging.
Multi-cancer
Reads signals from the 10 most common cancers in a single, unified platform.
Lab-ready
Uses standard handling, storage, and lab equipment, fitting into existing workflows and supports cost-effective testing at scale.
How it works
From sample to signal
The test turns a standard urine sample into a chemical fingerprint, then uses AI to decide whether a cancer signal is present and where it may be coming from.
Collect
A patient provides a urine sample—at home, in the clinic, or via mobile collection—using a simple kit and instructions.
Measure
A lab processes the sample using established methods similar to high-volume urine testing and measures thousands of volatile organic compounds (VOCs) and related features.
Analyze & report
AI models analyze the pattern of signals to determine whether a cancer signal is likely present and to rank likely cancer types. The goal is to return a clear, clinician-friendly result that can fit into existing workflows.
Performance & validation
Early performance and ongoing validation
Studies using real patient samples show encouraging performance in detecting cancers at stages 1 and 2 across multiple cancer types. As validation continues, we’ll keep updating this snapshot with new data.*
Sensitivity & specificity
≥90% sensitivity, ≥99.5% specificity
Early internal data show ≥90% sensitivity and ≥99.5% specificity for cancer vs non-cancer classification, with strong separation (AUC = 1.00).
Evidence base
30+ VOC cancer studies
30+ peer-reviewed studies on urine VOC cancer detection, a 691-sample internal multi-cancer proof-of-concept, and an 84-sample external blinded prostate study at Duke all support the platform’s approach.
Regulatory milestone
FDA Breakthrough for MCED (top 10 cancers)
FDA Breakthrough Device Designation for a multi-cancer early detection test across the 10 most common cancers.
*Performance metrics are based on current internal retrospective analyses, an external blinded study, and published VOC literature. They reflect studies conducted for clinical research and development, and may change as additional data become available. The Toby Oncology test is currently used in clinical research and trials and is not yet broadly available for routine clinical use or population-wide screening.
ENCOURAGING PERFORMANCE
A review of 63 VOC cancer studies found pooled sensitivity of 79% and specificity of 89% for detecting cancer.
— JAMA Network
For clinicians
Clinical workflow
Tests are ordered for appropriate patients. Urine is collected at home with a simple kit or during a doctor visit, then sent to a Toby-enabled lab for analysis and a clear result.
-
1
ORDER
Test ordered for appropriate adults based on clinical judgment and guidelines.
-
2
COLLECT
Urine sample collected at home using a Toby kit or during a routine visit.
-
3
RUN
Sample processed and analyzed in a lab running the Toby Oncology assay.
-
4
REVIEW
Result returned to the clinician to inform follow-up and further evaluation.
For labs and life sciences
Lab, trial, and monitoring applications
The Toby Oncology platform generates multi-cancer signal data that labs and life sciences teams can use in real-world workflows, clinical studies, and exploratory trials.
Labs and diagnostic networks
Reference labs and diagnostic networks can run the assay on standard equipment and existing sample-handling workflows, making it practical to scale testing volume as programs grow and to support repeat testing.
Trials and research
Life sciences and pharma teams use urine-based multi-cancer signals to help enrich high-risk cohorts, explore how signals relate to treatment response, and add a non-invasive readout alongside existing biomarkers and imaging.
Longitudinal monitoring
Current and planned studies are assessing how repeated urine samples can support longitudinal monitoring in high-risk patients and survivors, tracking changes in signals over time to inform follow-up.
For inquiries about lab deployments, clinical trials, or partnerships, please contact our team.
Under the hood
Technology and IP
Behind the test is a combination of chemistry, machine learning, and IP designed to be durable over time.
Signal and models
The platform reads thousands of VOC and related features per sample and uses machine learning to distinguish cancer from non-cancer and to localize likely cancer types.
Indications
Current studies span 10 major cancers, including both cancers with established screening programs and those with limited or no routine screening today.
IP MOAT
Patent filings cover key elements of sample processing, model design, biomarker combinations, and clinical indications, with the goal of protecting the performance “core” of the platform as it scales.
Next Steps
If you believe early detection should truly be early, we’d love to hear from you
For partnership or investment inquiries, please include your organization and area of interest.