General Biology
About us
The Precision Prevention Company
Chronic disease starts in the body long before anyone gets a diagnosis. By the time symptoms set in, care can only slow it down. Each of us has a biological fingerprint that shapes our risk and how we respond to treatment. Yet medicine still waits for illness and treats people as averages.
Most tools look at one signal at a time, like a gene or a blood test. That works when a disease has a single cause. But chronic disease rarely does, and the clues to who is at risk, how illness develops, and what will help sit across layers of biology rather than in any one test. General Biology, an MIT and Broad Institute spinout, builds AI foundation models that read those layers together. We use them to map how disease works, find preventive therapies and combinations, and predict who will respond, starting with Alzheimer's.
Why start with Alzheimer's? It is probably the hardest test. Changes in the brain begin up to two decades before memory fades. General Biology grew out of FINGERPRINT, a research effort seeded by MIT's Aging Brain Initiative at the Picower Institute with teams from the Broad Institute, Karolinska Institutet, and the FINGERS Brain Health Institute. Within months, our team built FINGERS-7B, the first AI foundation model for Alzheimer's prevention. It found new early biomarkers and showed it can tell who responds to preventive care.
Now we are taking that work from discovery into new therapies, launching with support from the Alzheimer's Disease Data Initiative. A version of FINGERS-7B will be open to researchers on the AD Workbench. And our models do not stop at the brain. When we link AI, human biology, and trial data, we can see how chronic disease develops and what shifts its course. Disease becomes preventable when we can catch it early enough to act.
Built for teams developing and testing therapies.
Every clinical study faces the same question: who does a therapy help, and why? Pharma, biotech, and academic groups all run into it. Our models show them who benefits and what to test next.
One model across layers of biology.
FINGERS-7B is a 7-billion-parameter model trained on genomic, proteomic, clinical, and lifestyle data from tens of thousands of people in the FINGER study and World-Wide FINGERS Network. It connects those layers to show who is at risk and who responds to an intervention.
The average patient doesn't exist.
Clinical trials report one average effect for everyone enrolled. If a therapy helps part of a study and does nothing for the rest, the average can fall near zero. A drug that worked gets shelved.
Get in touch
Read about how our research originated at the Broad Institute and MIT
