Testing
Generate repeatable CGM series and scenario fixtures with @glucoseiq/testing.
@glucoseiq/testing produces repeatable CGM data for unit tests, Storybook
stories, product demos, screenshots, and documentation. Give it the same
options and seed, and it returns the same chronological readings every time.
npm install --save-dev @glucoseiq/testingGenerate a series
import { generateCGMSeries } from '@glucoseiq/testing'
const readings = generateCGMSeries({
days: 14,
seed: 7,
start: '2025-01-01T00:00:00Z',
})
const firstReading = readings.at(0)
if (!firstReading) throw new Error('Expected generated readings')
firstReading
// {
// value: 96,
// unit: 'mg/dL',
// timestamp: '2025-01-01T00:00:00.000Z'
// }The generator combines a circadian baseline, three default meal responses, seeded noise, and optional nocturnal hypoglycemia. Values are clamped to 40–400 mg/dL before output conversion.
| Option | Default | Purpose |
|---|---|---|
days | 1 | Number of days to generate |
intervalMin | 5 | Minutes between readings |
seed | 42 | Pseudorandom seed |
start | 2024-01-01T00:00:00Z | ISO timestamp for the first reading |
basal | 110 | Baseline in mg/dL |
mealTimes | 07:00, 13:00, 19:00 | Meal times as minutes from midnight |
mealAmplitude | 70 | Peak meal-excursion amplitude in mg/dL |
noise | 8 | Noise amplitude in mg/dL |
nocturnalHypoDays | [] | Zero-based day indexes with a 02:00–04:00 dip |
unit | mg/dL | Output as mg/dL or mmol/L |
Test an analysis
Generated readings use the GlucoseReading contract from @glucoseiq/core,
so they can go directly into reading-based analytics such as analyzeGlucose:
import { analyzeGlucose } from '@glucoseiq/core'
import { generateCGMSeries } from '@glucoseiq/testing'
const readings = generateCGMSeries({ days: 14, seed: 19 })
const report = analyzeGlucose(readings)
if (!report.valid || !report.agpProfile) {
throw new Error('Expected a valid analysis with an AGP-style profile')
}
if (report.dataSufficiency.totalReadings !== 14 * 288) {
throw new Error('Unexpected generated reading count')
}
if (report.agpProfile.bins.length !== 288) {
throw new Error('Unexpected percentile-bin count')
}Use a fixed start in tests that group by calendar date or local time. The
default starts at midnight UTC.
Start from a scenario
The scenarios collection provides four named fixtures:
import { scenarios } from '@glucoseiq/testing'
const steady = scenarios.steadyDay()
const hypo = scenarios.hypoNight()
const volatile = scenarios.rollercoaster()
const withDropout = scenarios.gappyTrace()| Scenario | Shape |
|---|---|
steadyDay() | A calm, mostly in-range day with smaller meals and less noise |
hypoNight() | A day with a 02:00–04:00 nocturnal dip |
rollercoaster() | Larger post-meal excursions and more noise |
gappyTrace() | A day with readings removed from 15:00 through 16:30 UTC |
Scenarios return fresh arrays, so a test can safely derive a variant without changing the next fixture:
import { scenarios } from '@glucoseiq/testing'
const sparseHypo = scenarios.hypoNight().filter((_, index) => index % 2 === 0)Produce mmol/L fixtures
import { generateCGMSeries } from '@glucoseiq/testing'
const readings = generateCGMSeries({
unit: 'mmol/L',
days: 2,
seed: 23,
})
const firstReading = readings.at(0)
if (!firstReading) throw new Error('Expected generated readings')
firstReading.unit // 'mmol/L'The underlying curve is generated in mg/dL and converted to one decimal place for mmol/L output.
Synthetic data is not clinical validation
These fixtures are designed for repeatable software testing and visual demos. They do not model a specific person, sensor, therapy, or clinical outcome, and they are not a substitute for validated reference datasets.