Synthetic data only. Never enter patient information; nothing is sent or stored. For clinicians. Not FDA cleared.For licensed clinicians. Educational demonstration on synthetic data. Never enter real patient identifiers or protected health information: everything runs in your browser and nothing is sent or stored. Clinical decision support only; not FDA cleared; physician review is required.
CPE-002 · version 1.0.0
Metabolic Pattern Engine: prediabetes, metabolic syndrome and insulin-resistance risk, without overclaiming
Guideline criteria for glycemia and metabolic syndrome. Insulin resistance itself is never diagnosed from one number.
The Metabolic Pattern Engine separates what guidelines can classify from what they cannot. Prediabetes, diabetes-range findings and metabolic syndrome have formal criteria, so the engine applies them exactly. Insulin resistance has no such criterion for a single lab value, so the engine will not diagnose it.
Fasting insulin, HOMA-IR and the TG/HDL ratio are calculated and shown. They do not classify anything: their threshold rules are sandboxed until they have an assay-specific, externally validated profile.
Playground
Load a synthetic case or type your own values. The result recomputes as you type, and every rule links to its page.
Glycemic testsDrag the sliders to cross a threshold
- <100
- 100–125 prediabetes
- ≥126 diabetes range
- <5.7
- 5.7–6.4 prediabetes
- ≥6.5 diabetes range
75-g protocol
Lipids, body and blood pressure
- <150
- 150–499
- ≥500 priority
- ≥1000 urgent
Metabolic syndrome components0 of 5
- ◌ Waist not met
- ◌ Triglycerides not met
- ◌ Low HDL-C not met
- ◌ Blood pressure not met
- ◌ Fasting glucose not met
● met · ◌ not met · ○ not entered. Three or more meet the criteria; that does not confirm insulin resistance.
Patient and scope
Symptoms and safety
Risk context
Liver context
Medications
Outside the standard adult runtime
Fasting insulin (supporting only)Calculation and sandbox
Shown with HOMA-IR when fasting and same-collection are confirmed. It does not classify anything.
Research sandbox (unvalidated)
In production the insulin and HOMA-IR rules are off: no validated threshold profile exists. Here you can supply your own threshold to see how the sandbox rules would behave. The output is labeled a sandbox candidate and is never a diagnosis.
What it does
- Classifies prediabetes by confirmed-fasting glucose, HbA1c or a protocol-confirmed 75-g OGTT.
- Flags diabetes-range findings for physician review, and checks the type (possible T1D/LADA) before anything else.
- Counts the five metabolic-syndrome components with sex- and population-specific thresholds.
- Surfaces screening opportunities from risk context (acanthosis, family history, prior gestational diabetes, PCOS, atherogenic dyslipidemia).
- Adds validity modifiers: fasting not confirmed, HbA1c interference, treatment, steroids, insulin therapy.
What it never does
- Confirm insulin resistance from fasting insulin, HOMA-IR or TG/HDL alone.
- Confirm MASLD from ALT alone.
- Assign a numeric confidence score or a symptom score.
- Recommend a medication or a dose.
Safety gates
These gates run before the pattern is shown, and urgent findings are displayed first.
- Pregnancy, pediatrics, known type 1 diabetes, hyperglycemic emergency and MODY are routed out of the adult runtime.
- Diabetes-range glucose with weight loss, ketones or classic symptoms raises an urgent review of diabetes type and suppresses insulin-resistance framing.
- Low glucose with inappropriate insulin raises an urgent hypoglycemia-pathway review.
- Triglycerides ≥500 and ≥1000 mg/dL raise priority and urgent pancreatitis-risk reviews.
Inputs
- Fasting glucose, HbA1c, 2-hour OGTT, random glucose
- Triglycerides, HDL-C, waist, blood pressure
- Fasting insulin (supporting only)
- Fasting status and protocol confirmations
- Medications, pregnancy and risk modifiers
Scope
In scope
- Adult, non-pregnant, ambulatory patients
Routed out
- Pregnancy
- Pediatrics or puberty
- Known type 1 diabetes (primary management)
- Hyperglycemic emergency
- Known MODY (primary management)
The full rule catalog
37 active rules and 4 sandbox or context entries. Each has its own page with conditions, outputs and evidence. 37 of 37 active rules are exercised by at least one synthetic case.
Validity and context
Safety alerts
Glycemic state
Metabolic syndrome
Lipid patterns
Clinical modifiers
Cross-engine triggers
Missing data
Insulin-resistance–associated pattern
Longitudinal (sandbox)
Synthetic cases
Every case is synthetic. Open one to see the inputs, the full result and the rule trace; then change the values.
CPE002-FX-001
Healthy Metabolic Baseline
A 35-year-old with normal glucose, lipids, waist and blood pressure.
CPE002-FX-002
Prediabetes by HbA1c
HbA1c of 6.0% and nothing else.
CPE002-FX-003
Prediabetes by FPG
Confirmed-fasting glucose of 108 mg/dL.
CPE002-FX-004
OGTT-Only Prediabetes
Only a 75-g OGTT: 2-hour glucose 165 mg/dL.
CPE002-FX-005
Diabetes-Range Finding
HbA1c 6.8% and fasting glucose 134 mg/dL.
CPE002-FX-006
Metabolic Syndrome with Normal Glucose
Normal glucose, but waist 108 cm, TG 190, HDL 36 and BP 138/88.
CPE002-FX-007
Hyperinsulinemic Pattern with Normal Glycemic Tests
Normal glucose and HbA1c, fasting insulin 22 µIU/mL, TG 170, HDL 42.
CPE002-FX-008
Acanthosis with Insufficient Lab Data
Acanthosis nigricans on exam. No labs yet.
CPE002-FX-009
Vague Symptoms without Objective Findings
Postprandial fatigue, brain fog and carbohydrate craving, with normal labs.
CPE002-FX-010
Treated State with Metformin and GLP-1
Prediabetes labs in a patient on metformin and a GLP-1 agonist.
CPE002-FX-011
Steroid-Associated Dysglycemia Context
Fasting glucose 112 mg/dL during a prednisone course.
CPE002-FX-012
Endurance Athlete with FPG Prediabetes Criterion
A marathon runner with fasting glucose 104 mg/dL.
CPE002-FX-013
Possible T1D or LADA
Random glucose 380 mg/dL, weight loss, thirst and positive ketones.
CPE002-FX-014
Hypoglycemia with Inappropriate Insulin
A low-glucose event with inappropriately high insulin.
CPE002-FX-015
Severe Hypertriglyceridemia
Triglycerides of 640 mg/dL.
CPE002-FX-016
Very Severe Hypertriglyceridemia
Triglycerides of 1,240 mg/dL.
CPE002-FX-017
HbA1c Potentially Unreliable
HbA1c 6.0% in a patient with a known hemoglobin variant.
CPE002-FX-018
Pregnancy Out of Standard Runtime
A positive pregnancy test and a glucose of 98 mg/dL.
CPE002-FX-019
MASLD Cross-Trigger with FIB-4
Hepatic steatosis on ultrasound. AST 48, ALT 62, platelets 170.
CPE002-FX-020
Longitudinal Metabolic Deterioration
Six months of rising HbA1c, triglycerides and waist.
CPE002-FX-021
Metabolic Syndrome without Insulin Measurement
Waist 96 cm, TG 165, HDL 44 and no insulin measurement.
CPE002-EX-01 · extended
Fasting insulin drawn without confirmed fasting
A fasting insulin of 18 µIU/mL and glucose 105 mg/dL, but the patient was not fasting.
CPE002-EX-02 · extended
Patient on insulin therapy with a high insulin level
Insulin 30 µIU/mL in a patient who injects insulin, with fasting glucose 150 mg/dL.
CPE002-EX-03 · extended
Family history, prior gestational diabetes and PCOS with normal glucose
Normal fasting glucose, but three high-risk modifiers.
CPE002-EX-04 · extended
Normal fasting glucose but HbA1c 6.0%
Two glycemic tests that disagree.
CPE002-EX-05 · extended
Medication list unavailable
Fasting glucose 108 mg/dL and no medication history on file.
Evidence
Guidelines and references the rules cite. Each rule page lists the ones that support it.
EV-ADA-2026-DIAGAmerican Diabetes Association Professional Practice Committee for Diabetes. Diagnosis and Classification of Diabetes: Standards of Care in Diabetes—2026. Diabetes Care. 2026;49(Suppl 1):S27-S49.
Supports: Prediabetes criteria based on A1C, fasting plasma glucose, and 2-hour plasma glucose after 75-g OGTT. Diabetes-range findings based on A1C, fasting plasma glucose, OGTT, or symptomatic random plasma glucose.
EV-NIDDK-IRNational Institute of Diabetes and Digestive and Kidney Diseases. Insulin Resistance & Prediabetes.
Supports: Insulin resistance testing is not routine diagnostic testing in the same way as prediabetes testing. Prediabetes testing uses A1C, fasting plasma glucose, or OGTT.
EV-METS-2009IDF/AHA/NHLBI/WHF/IAS/IASO Joint Interim Statement. Harmonizing the Metabolic Syndrome. Circulation. 2009;120:1640-1645.
Supports: Metabolic syndrome is classified when three abnormal findings out of five are present. Waist thresholds require population-specific configuration.
EV-NGSP-A1CNational Glycohemoglobin Standardization Program. Factors that Interfere with HbA1c Test Results.
Supports: Hemoglobin variants and altered red-cell lifespan may interfere with A1C interpretation. Iron deficiency may cause higher A1C values.
EV-ACC-TG-2021American College of Cardiology Expert Consensus Decision Pathway. Persistent Hypertriglyceridemia. 2021.
Supports: Triglycerides >=500 mg/dL require severe hypertriglyceridemia review. Triglycerides >=1000 mg/dL warrant urgent attention for pancreatitis/chylomicronemia risk.
EV-AASLD-MASLDAmerican Association for the Study of Liver Diseases. Noninvasive Assessment of Patients with MASLD.
Supports: FIB-4 <1.3 is generally used as an initial low-risk result in appropriate populations. FIB-4 >=1.3 prompts consideration of secondary assessment within a hepatic pathway.
Frequently asked questions
Can this engine diagnose insulin resistance?
No. Fasting insulin, HOMA-IR and the TG/HDL ratio are shown as calculations only. Their interpretation rules are sandbox-disabled until an assay-specific threshold profile and external endocrinology validation exist. The engine reports risk context and screening opportunities instead.
Which criteria does it use for prediabetes and diabetes?
The ADA Standards of Care diagnostic criteria: fasting plasma glucose 100–125 mg/dL, HbA1c 5.7–6.4%, or 2-hour OGTT 140–199 mg/dL for prediabetes; 126, 6.5% and 200 mg/dL for the diabetes range.
What happens when HbA1c may be unreliable?
A documented hemoglobin variant, recent transfusion, hemolysis, iron deficiency, recent bleeding or similar interference removes HbA1c as a basis for classification and limits the interpretation.
How is metabolic syndrome counted?
Three of five components: elevated waist (population-specific), triglycerides ≥150 mg/dL or treated, low HDL-C by sex, blood pressure ≥130/85 mmHg or treated, and fasting glucose ≥100 mg/dL or treated.
Why does an athlete or a patient on steroids still get the same criteria?
Context modifiers are shown beside the finding. They do not erase an ADA-defined glycemic criterion, and a steroid course explicitly stops the engine attributing dysglycemia to baseline insulin resistance alone.
Specification status: Specification v1.0.0. Sandbox rules are disabled by default. Reviewed by Dr. Vinicius Amaral, D.O. Not for clinical production use.
Clinical review: Dr. Vinicius Amaral, D.O., Medical reviewer, 2026-10-03.
Keep going
CPE-005
MASLD Pattern Engine
Steatosis must be documented before MASLD is named. Fibrosis risk follows FIB-4, then elastography or ELF.
CPE-007
Cardiovascular Risk Pattern Engine
ACC/AHA categories applied to a risk you enter, with risk enhancers, Lp(a), apoB, calcium score and the 2022 LDL-C thresholds.
CPE-010
PCOS Pattern Engine
Two of three criteria in adults, irregular cycles plus hyperandrogenism in adolescents, and PCOS only after other causes are excluded.
Resources
Micronutrient patterns and cognitive dysfunction
Game
Pattern challenge
Guess what the engine will report.
Try Biod in your practice
The Clinical Intelligence Layer — transforming fragmented clinical information into structured intelligence for clinician review and action.