Bilirubin/Jaundice Risk Reference (education) mechanism

Model scope

NEEDS_REVIEW because the canonical curve dataset, interpolation method, validated age and gestational-age scope, unit contract, and mapping from Bhutani's four zones to the requested three labels are not specified. The existing educational-only disclaimer is appropriate but should explicitly state that the display is not a diagnostic tool and not a phototherapy or exchange-transfusion threshold.

• The Bhutani nomogram is a lookup against percentile curves, but the specification does not provide the actual curve table or define whether age interpolation is linear, spline-based, or nearest-row lookup.
• The Bhutani nomogram conventionally distinguishes four zones: low-risk (<40th percentile), low-intermediate-risk (40th–75th), high-intermediate-risk (75th–95th), and high-risk (>95th). The stated three outputs—low/intermediate/high—do not define how the four zones are collapsed.
• Boundary behavior is unspecified. In the vectors below, the educational mapping assumes low = below the 40th percentile, intermediate = 40th through 95th percentile inclusive, and high = above the 95th percentile.
• The nomogram is not a treatment-threshold calculator and must not be used to determine phototherapy or exchange-transfusion treatment.
• Applicability constraints are missing: the classic Bhutani nomogram is generally used for infants at or above 35 weeks' gestation, during its validated postnatal-age range, commonly approximately 18–144 hours. Gestational age and clinical context are absent inputs.
• The specification does not distinguish total serum bilirubin from transcutaneous bilirubin, direct bilirubin, or indirect bilirubin.

Units and conventions

• Bilirubin values in the reference table are total serum bilirubin in mg/dL.
• For the conversion-trap vector, 1 mg/dL is treated as 17.1 micromol/L; therefore 171 micromol/L = 10.0 mg/dL.
• Age is postnatal age in completed hours, not gestational age or day-of-life rounded to 24-hour blocks.
• The test vectors use commonly reproduced, rounded Bhutani nomogram anchor values. A production implementation requires a validated, explicitly sourced digitization of the chosen Bhutani curve.

Inspectable vector

One of 4 source-bound review vectors:

{
  "inputs": {
    "age_hours": 24,
    "total_serum_bilirubin_mg_dL": 5
  },
  "expected_outputs": {
    "percentile_zone": "intermediate",
    "boundary_interpretation": "exactly at 40th percentile"
  },
  "reference_data": {
    "age_24_hours": {
      "5th_percentile_mg_dL": 3,
      "40th_percentile_mg_dL": 5,
      "75th_percentile_mg_dL": 6,
      "95th_percentile_mg_dL": 8,
      "source_convention": "Bhutani et al., Pediatrics 1999, commonly reproduced rounded nomogram anchor values"
    }
  },
  "working": "At 24 hours, the 40th-percentile boundary is 5.0 mg/dL. Input = 5.0 mg/dL. 5.0 >= 5.0, so it is not low. It is also 5.0 <= 8.0, the 95th-percentile boundary. Under the stated boundary convention, the result is intermediate."
}

Failure boundary

BREAK_CLAIMED · SELF_FLAGGED

The instrument is specified as a lookup-table tool, but the required Bhutani nomogram data points, validity interval, interpolation policy, risk-zone thresholds, and input units are not supplied. It therefore cannot produce a defensible percentile zone for ordinary or boundary inputs and must reject or explicitly flag them rather than inventing results.

Other records

Sources and evidence

These pointers bind the page to evidence. They do not grant authority or collapse verification into deployment.

• VERIFIED_LEDGER.json
• evidence-tranche7/tranche7-results.json
• BOUNDARY_HAZARD_CATALOG.json