AI FOR SCIENCE

AI that decodes biological language from blood to detect cancer early.

MicroTrace BioAI develops fragment-native AI to decode blood’s molecular signals.

Explore the science →Work with us →

Currently in research and development; not yet available for clinical use.

WHY PANCREATIC CANCER

One of the deadliest cancers. No definitive blood test to resolve diagnostic uncertainty.

Pancreatic cancer is often found too late for curative surgery, while benign lesions can lead to unnecessary invasive procedures. When imaging is inconclusive, clinicians must choose between surveillance, biopsy, and surgery—with limited biomarker support, as benign conditions can also elevate CA19-9.

~80%

of pancreatic cancers are diagnosed at a stage where surgery is no longer curative

SEER, 2025
~13%

five-year relative survival, all stages

American Cancer Society, 2025
72% / 86%

pooled sensitivity and specificity of CA19-9 across 79 studies

Zhao et al., 2022

THE INSIGHT

Every molecule carries more than one signal. Fragment-native AI learns them together.

Blood carries complementary signals across DNA, RNA, and proteins. Our multimodal AI approach aims to learn from this molecular richness, preserving information that summary statistics can miss to uncover patterns relevant to cancer.

Same molecule, multiple signals

Encode multi-layer molecular signals together on each fragment.

Learn before compressing

Models learn molecular representations before reducing the signal.

Built for sparse signal

The rarest signals matter most, so our models are designed for biological sparsity.

HOW WE WORK

Recursive scientific intelligence

Recursive Science Intelligence turns every experiment into a better next decision.

FIRST PROGRAM

A blood-based diagnostic aid for indeterminate imaging findings.

Imaging can identify an abnormality without fully resolving malignancy risk. Our first program is a plasma-based classifier intended to add information at that decision point.

See the program →
Indeterminate imaging
→
Blood draw
→
Risk stratification
→
Escalate or continue surveillance

RESEARCH

Computational feasibility demonstrated.

A genomic foundation model fine-tuned on sparse cell-free DNA methylation data, outperforms conventional machine-learning baselines in gastrointestinal cancer classification.

Read the research →

THE TEAM

A team of scientists with industry experience in developing and commercializing cancer diagnostics.

MicroTrace brings together PhD scientists and a physician-scientist with expertise in cancer biology, AI, and clinical research. Our experience spans cancer test development and commercialization, multi-omic assay R&D, and large-scale machine learning.

Clinical assay R&DFoundation models for genomicsStudy designCancer genomicsClinical product developmentLarge-scale ML systems

COLLABORATE

We welcome collaborations with clinicians, researchers, and industry partners.

We are building the study, assay, and clinical workflow with partners who work directly with molecular data, and cancer diagnostics.

Start a conversation →