Tools · General business
Absolute Risk Translator
Turn a scary relative-risk headline into the actual absolute risk, before and after.
How it works
Deterministic risk math (compute.ts) OVER a corpus-grounded framing. Give it a baseline rate + a relative multiplier ('doubles' = 2), and code computes the absolute risk before/after, the absolute change in percentage points and extra cases, and number-needed-to-harm — the LLM never touches a number, it only writes an evidence-first plain-language framing grounded in the pregnancy corpus (Expecting Better's absolute-over-relative discipline). For The Family Almanac's pregnancy guide.
You bring
{ outcome, claim?, baseline_count, baseline_per, relative_multiplier, cluster?, model? }
You get
{ baseline, adjusted, absolute_change_pp, change_per_baseline, number_needed, framing, worth_worrying, grounded_in, provenance }
Use it for
- →'Coffee doubles miscarriage risk' + baseline 100/1000 → '100 in 1,000 → 200 in 1,000' with an honest is-this-worth-worrying read
- →Show a parent that a scary-sounding doubling of a tiny risk is still tiny (or that a big baseline is different)
Run it on your data
Call it on your own inputs — over the API, or hand it to your AI agent via MCP. Discovery is open; running it is metered.