Decode bone marrow to improve human health
Millions of people develop disease each year because of what happens in their bone marrow.

Blood cancers. Autoimmune diseases. TB. Sepsis. Cardiovascular disease. When solid tumors kill, it's often by co-opting the bone marrow. Mutations there drive aging. It's a highly complex system.
The Central Role of Bone Marrow
Your bone marrow makes 90% of all cells your body replaces. It's where your immune system develops and matures, making it one of the most information-rich tissues in our bodies.
Recent research has mapped millions of single cells in healthy bone marrow, revealing dozens of distinct cellular neighborhoods that control stem cell fate. The spatial architecture matters—where cells sit, who they talk to, how they mature drives disease.
For example, spatial disruption has been linked to therapy failure in relapsed myeloma, and specific immune cell neighborhoods can predict both survival and neurotoxicity after CAR-T treatment.
The Gap
Whether due to hematology's specialist nature or clinical fragmentation, this potentially life-saving data remains painfully underexplored.
Bone marrow is effectively absent from major public datasets. Clinical molecular analysis tools are like trying to understand a city only through population counts, limited in scope and unable to capture complexity. Pathologists have incredible intuition—they can sometimes guess mutations by sight—but they miss patterns invisible to the human eye.
The gap isn't just technological. Platforms today fragment attention across multiple tissues, which works for digitizing current approaches. But discovering novel signals requires focus to shift from incremental to transformative insights.
Our Approach
We obsess over bone marrow. Single-tissue focus means our models' knowledge compounds with every sample. What we learn from leukaemia helps us understand an autoimmune disease—it's all fundamental bone marrow health.
The core breakthrough won't be just better algorithms—it's the dataset that matters most. The right data enables us to extract molecular signatures from tissue images: gene expression patterns, protein levels, and treatment response markers that match or outperform conventional approaches.
Why does this work? Tissue contains, encapsulates, and reflects many layers of information at different scales—expressing what lies beneath its surface. Biologically, tissue images capture what other data types miss: genes, proteins, and spatial relationships all in one.
The Leap
The key advance isn't digitizing established markers. It's combining differentiated data and deep learning to find patterns we'd never think to look for.
We analyze the images of bone marrow every hospital already takes. Our models aim to predict who will respond to treatment, whose disease will progress, and which patients face hidden risks.
Starting with blood cancers, we're uncovering connections between bone marrow and broader health: its role in age-related changes, autoimmune conditions, and other diseases where bone marrow involvement is still being understood.
This is a unique platform at the tissue level—the right biological scale to decode bone marrow and understand its role in human health.
Join us in decoding bone marrow.