Framingham-style Cardiovascular Risk Score (education) mechanism

Model scope

Formula verification is conditional because the specification says 'Framingham-style' and simultaneously references Framingham and ACC/AHA pooled-cohort equations. These are not interchangeable. The four vectors use the 2008 sex-specific Framingham general CVD equation with natural-log predictors, sex-specific mean sums, and baseline survival constants. Before implementation, select and identify exactly one equation, endpoint, coefficient edition, required demographic variables, treatment definition, valid age range, unit system, and category thresholds.

• The specification does not identify one unique published equation: Framingham general cardiovascular disease, Framingham hard coronary heart disease, and ACC/AHA pooled-cohort ASCVD equations have different coefficients, endpoints, calibration constants, and required inputs.
• The stated inputs are insufficient for the ACC/AHA pooled-cohort equation because race and antihypertensive-treatment status are required; treatment status is also required by the selected Framingham general CVD coefficient set.
• The output risk category is underspecified because category thresholds are not supplied and differ among guidelines and endpoints.
• The selected test convention is the sex-specific 2008 Framingham general cardiovascular disease equation, not an ASCVD pooled-cohort equation. The implementation must explicitly label this endpoint and equation.

Units and conventions

• Age is in completed years.
• Total cholesterol, HDL cholesterol, and systolic blood pressure use the conventional Framingham units of mg/dL, mg/dL, and mmHg, respectively.
• The equation uses natural logarithms of age, total cholesterol, HDL, and systolic blood pressure.
• Smoking and diabetes are binary indicators: 1=yes and 0=no.
• Systolic-blood-pressure coefficients differ for untreated and treated hypertension; treatment status is therefore required.
• The illustrative risk-category convention used in these vectors is low <10%, intermediate 10% to <20%, and high >=20%; this is an assumed convention, not specified by the instrument.

Inspectable vector

One of 4 source-bound review vectors:

{
  "inputs": {
    "age_years": 50,
    "sex": "male",
    "total_cholesterol_mg_dL": 200,
    "hdl_mg_dL": 50,
    "systolic_bp_mmHg": 120,
    "bp_treated": false,
    "smoker": false,
    "diabetes": false
  },
  "expected_outputs": {
    "risk_10yr_percent": 7.24,
    "risk_category": "low"
  },
  "reference_data": {
    "equation": "Framingham 2008 general CVD, men",
    "source_convention": "widely tabulated standard value (D'Agostino et al., Circulation 2008)",
    "coefficients": {
      "ln_age": 3.06117,
      "ln_total_cholesterol": 1.1237,
      "ln_hdl": -0.93263,
      "ln_sbp_untreated": 1.93303,
      "smoker": 0.65451,
      "diabetes": 0.57367
    },
    "mean_sum": 23.9802,
    "baseline_survival": 0.88936,
    "clinical_benchmark_values": {
      "total_cholesterol_mg_dL": 200,
      "hdl_mg_dL": 50,
      "systolic_bp_mmHg": 120,
      "source_convention": "widely tabulated standard clinical benchmark values"
    },
    "risk_category_thresholds": {
      "low": "<10%",
      "intermediate": "10% to <20%",
      "high": ">=20%",
      "source_convention": "assumed illustrative convention, not part of the published equation"
    }
  },
  "working": "ln(50)=3.912023; ln(200)=5.298317; ln(50)=3.912023; ln(120)=4.787492. Linear predictor = 3.06117(3.912023) + 1.12370(5.298317) - 0.93263(3.912023) + 1.93303(4.787492) + 0 + 0 = 11.975367 + 5.953719 - 3.648470 + 9.254366 = 23.534982. Difference from mean = 23.534982 - 23.9802 = -0.445218. exp(-0.445218)=0.64070. Risk = 1 - 0.88936^0.64070 = 1 - 0.92760 = 0.07240 = 7.24%."
}

Failure boundary

BREAK_CLAIMED · SELF_FLAGGED

The instrument cannot produce reproducible numerical outputs from the supplied specification: it does not identify one exact Framingham or ASCVD equation, provide coefficient and baseline-survival tables, define valid input ranges, specify units, or define risk-category thresholds. It also has undefined behavior for domain-invalid values such as HDL = 0.

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