How a plan is built.
Preventah holds a catalog of 117 conditions and a curated library of prevention guidance. Neither one names the other. They meet through a small, closed vocabulary of tags.
Conditions carry tags
High blood pressure carries diet-salt, activity, weight, alcohol, sleep and stress. The tags describe what is worth doing, never how likely anything is.
Guidance carries the same tags
An item about keeping salt under a teaspoon carries diet-salt. When your tags and an item’s tags overlap, you see the item. That is the whole matching rule.
What this design buys
117 conditions share one body of guidance, so adding a condition is a data change rather than a writing job, and nobody has to invent a bespoke paragraph about a condition they are not qualified to write about.
Some conditions carry no dietary or exercise tag at all — an inherited retinal condition, for instance. Those fall back to general prevention rather than inventing a connection that is not there.
What Preventah refuses to do.
No scoring or prediction
Family-history relevance is used to word a sentence, never to compute a risk number. Nothing tells you how likely you are to develop anything.
No live AI
Resolving a plan is a pure function of your selections and the day. No model call, nothing asynchronous, and the same inputs always produce the same plan.
No targets
Measurement trends report which way a number moved and by how much. They never say whether that is good, and no reading is ever colour-coded.
No diagnosis or prescription
Several items carry a safety note telling you to skip them. Waist measurement is the wrong habit to hand someone with a history of disordered eating, and fibre is not universally good for every gut.
Where the guidance comes from.
Public-health references: the World Health Organization, the NHS, the CDC and MedlinePlus. They are given so you can read further, not as a citation for a numeric claim, because Preventah makes no numeric claims.