AGP & Rendering

The AGP series

buildAGPProfile pools readings by minute-of-day (time-zone aware) across all days and reduces each bin to percentile bands — the exact data behind the AGP chart clinicians read:

import { buildAGPProfile } from '@glucoseiq/core'

const profile = buildAGPProfile(readings, {
  timeZone: 'America/New_York', // local clock, DST-correct
  binMinutes: 5,                // 288 bins/day
  percentiles: [5, 25, 50, 75, 95],
})

profile.bins[96]
// { minuteOfDay: 480, percentiles: { 5: 92, 25: 104, 50: 118, 75: 134, 95: 161 }, n: 14 }

Draw it with any chart library as two shaded areas (5–95, 25–75) plus a median line — or don't draw it yourself at all:

AGP-in-a-tag

import { agpChartToSVG } from '@glucoseiq/core'

const svg = agpChartToSVG(readings, { theme: 'dark', title: 'Last 14 days' })
// A complete, self-contained <svg> string — no DOM, no canvas, no framework.

It renders anywhere a string does: innerHTML, email, PDF, a GitHub README, React Server Components. Companions: tirBarToSVG (the five-zone stacked bar) and trendTileToSVG (the glanceable current-value tile).

Reproducibility

Nearest-rank percentiles are the default — matching glucosePercentiles and keeping cited results reproducible. Interpolated (type-7) percentiles are an explicit opt-in via method: 'linear'.