How it works + the science
First, we gather the story.
Then we do the maths.
Your health is not one appointment. So, why should a risk assessment behave as if it is?
The process
Four steps.
No magic tricks.
Automation can do a great deal. It can find a lab result on page 47. It can explain a complicated output. It cannot quietly invent a clinical input or calculate a risk score.
Bring the pieces together
Connect a supported health record, upload a document, or type in what you know. Every value keeps its date and source.
Check the record
A review agent may suggest a medication, diagnosis, or measurement. You decide whether it is right before anything is used.
Run the right model
Separate, versioned code calculates each disease risk. Language models do not calculate, revise, or round the score.
Make the result useful
The report explains the horizon, the factors that matter, what is missing, and the questions worth asking your clinician.
The scientific boundary
The score comes from the model.
Not the AI.
A report agent receives a read-only result. It may translate the output into normal language. It cannot reach back into the calculation and change the number. That separation is not a technical footnote. It is the point.
What do we actually know?
Clinical records, lab trends, medicines, family history, behaviour, environment, access, and the source of every value.
Which equation belongs here?
A published, disease-specific implementation with a stated population, time horizon, version, and limitations.
What can we reasonably say?
The estimate, its drivers, missing information, uncertainty, and questions for a qualified healthcare professional.
The disease programme
One platform.
Several separate questions.
Diabetes is not kidney disease. Stroke is not COPD. We refuse the temptation to flatten everything into one dramatic, meaningless wellness score.
01Type 2 diabetes02Cardiovascular disease03Heart failure04Chronic kidney disease05Stroke06COPD07Liver disease
A living source ledger
Trust, but make it clickable.
AHA PREVENT
10-year and 30-year cardiovascular risk, including cardiovascular, kidney, metabolic, and optional social context.
Read source ↗ARIC diabetes scoresPublished diabetes risk systems derived from a prospective cohort of 12,729 U.S. adults.
Read source ↗CDC kidney diseaseMore than 1 in 7 U.S. adults may have CKD, and as many as 9 in 10 do not know it.
Read source ↗These references document the model families and public-health evidence behind the platform. Each implementation retains its population, time horizon, version, and stated limits.