Plasma proteomics reads the functional state of every system in your body. We translate that signal into clinically actionable risk and trajectory, years before symptoms appear, across organ systems no single lab test can see.
Our blood plasma contains thousands of circulating proteins. Each one is a real-time signal of what is happening in a specific tissue: liver enzymes report on hepatic function, neuronal markers report on brain integrity, hormonal proteins report on the endocrine system.
Together, these proteins form a functional readout of every major organ system at once. Standard blood panels measure 30–60 markers. Kale measures more than 1,000.
Genomics tells you what's possible. Wearables tell you what's happening at the surface. Plasma proteomics tells you what's happening, right now, deep inside your biology, and where it's headed.
Proteins are tissue-specific. Liver proteins report on liver function. Neuronal proteins report on brain. Cardiac proteins report on heart. From a single blood draw, plasma proteomics captures the functional state of every major organ system at once, something no other clinical test can do.
Protein signatures change years to decades before clinical symptoms emerge. Amyloid-related proteins begin shifting 15–20 years before an Alzheimer's diagnosis. Cardiac stress proteins predict heart failure a decade out. This is the prevention window we are designed to act inside.
Unlike static genetic risk, protein signatures change in response to treatment, often within weeks. This makes proteomics the only modality that can both (a) predict the risk and (b) verify whether intervention is working, in the same person, across multiple time points.
A raw proteomic readout is unusable in a clinical context. The work, the actual technical work, is translating thousands of measurements into a handful of interpretable, actionable scores. Here's how.
We work with proteomic data from tens of thousands of individuals spanning healthy and disease cohorts. Statistical discovery identifies which combinations of proteins separate disease from non-disease, years before symptoms.
Machine learning compresses high-dimensional protein patterns into single interpretable scores. We weight proteins by their contribution to outcomes, regularize against overfitting, and explain every score with feature attribution.
Every signature is validated in independent cohorts before clinical use, then continuously monitored as new data accrues. We report performance honestly: what works, where it fails, where confidence is bounded. And update accordingly.
When Kale outputs a risk trajectory, this is what you're looking at, what it tells you, and — equally important — what it doesn't.
A trajectory plots the probability that someone with your proteomic signature would be diagnosed with a given condition by year N, alongside a population baseline. The gap between the two is the headroom for prevention.
Predictive horizons are validated signature by signature, not asserted universally. Cardiometabolic and inflammatory signals carry meaningfully a few years out; cognitive trajectories, on the strongest protein panels, extend further. AUC across our disease panel typically ranges from 0.7 to 0.9 — and we report it honestly per signature.
A trajectory is a probabilistic estimate, not a verdict. It tells you where your biology is heading without intervention — surfacing that early is the precondition for changing it. Every signature is reviewed by a licensed clinician before delivery.
Plasma proteomics is not a new idea. The last decade has produced a body of peer-reviewed evidence that makes the technology clinically defensible. Kale's signatures are built on this foundation, not invented from scratch.
Below: a curated set of the most consequential papers across organ aging, disease-risk prediction, and intervention-response. The science our platform is built on.
Want to discuss research collaborations? We're always open to engaging with partners.