Researchers at MIT and Microsoft have developed an AI system that designs molecular sensors capable of detecting cancer through a simple urine test. The technology, called CleaveNet, could eventually enable at-home screening for dozens of cancer types using a paper strip similar to a pregnancy test.
How CleaveNet Works
Cancer cells overproduce certain enzymes called proteases. CleaveNet designs short protein chains (peptides) that these cancer-specific proteases will cut. The system works by coating nanoparticles with these designer peptides. When injected, the nanoparticles travel through the body. If cancer-linked proteases are present, they cleave the peptides, and the resulting fragments get excreted in urine.
A paper strip test then detects these fragments, indicating which proteases are active and potentially which cancer type is present.
The research team, led by MIT professor Sangeeta Bhatia and Microsoft Research principal researcher Ava Amini, published their findings in Nature Communications on January 6, 2026.
The Numbers
The team claims their approach could “detect and distinguish between 30 different types of cancer” through an at-home diagnostic kit. They’ve already demonstrated sensors for lung, ovarian, and colon cancers in laboratory settings.
The AI system targets matrix metalloproteinases and other protease classes that are overactive in early-stage tumors, potentially catching cancer before symptoms appear.
What This Means
Early detection dramatically improves cancer survival rates. Pancreatic cancer caught at stage I has a 44% five-year survival rate; caught at stage IV, that drops to 3%. But most cancers lack simple screening tests, meaning they’re often found late.
A cheap, at-home urine test could change that calculus entirely. It wouldn’t replace imaging or biopsy for diagnosis, but it could serve as a first-line screening tool that flags when someone should see a doctor.
The Fine Print
This is still laboratory research. The team has demonstrated the concept works in controlled settings, but clinical trials in humans are years away. Several hurdles remain:
- The nanoparticles need to prove safe for human injection
- False positive and negative rates need to be established in diverse populations
- The system needs to distinguish cancer signals from other conditions that elevate protease activity
- Regulatory approval for both the nanoparticles and the diagnostic test
The research was funded by the La Caixa Foundation, Ludwig Center at MIT, and Marble Center for Cancer Nanomedicine. The team includes scientists from both academia and Microsoft Research, suggesting commercial interest in developing this technology further.
If successful, this would represent a shift in how we screen for cancer: from expensive imaging at medical facilities to routine at-home testing, similar to how blood glucose monitoring changed diabetes management.