Know your diabetes risk
and what moves it

What the numbers are, and aren't

Level 1: validated questionnaires

FINDRISC (Finnish Diabetes Risk Score) scores 8 items from 0 to 26. Each score band maps to the published estimate of 10-year type 2 diabetes risk from the original Finnish cohort: about 1 in 100, 1 in 25, 1 in 6, 1 in 3 or 1 in 2. It uses its own European BMI and waist cut-offs. IDRS (Indian Diabetes Risk Score) scores 4 items from 0 to 100 using Asian waist cut-offs. It was developed to screen for undiagnosed diabetes in India. Sleep and smoking are asked but are not part of either score. They only shape the guidance.

Levels 2–3: criteria plus an illustrative index

The glucose category uses ADA diagnostic thresholds: HbA1c 5.7–6.4% or fasting glucose 100–125 mg/dL is the prediabetes range; HbA1c ≥ 6.5% or fasting glucose ≥ 126 mg/dL is the diabetes range. A diagnosis always needs confirmation by a clinician, usually with repeat testing. The other indicators are:

• Metabolic-syndrome criteria: harmonised definition, any 3 of 5.
• HOMA-IR: an insulin-resistance estimate from fasting glucose and insulin.
• FIB-4: a liver-fibrosis index.
• eGFR: kidney function, calculated with the race-free CKD-EPI 2021 equation and staged by KDIGO.

The 0–100 index is a hand-weighted logistic formula. Its weights were chosen for illustration and were not fitted to or validated on any data, so it is not a probability. It exists to show how factors move together. Each bar shows weight × (your value − reference value). The reference person is 40 years old with BMI 23, waist at the cut-off, no family history, regular activity, fasting glucose 90 mg/dL, HbA1c 5.3% and systolic BP 120 mmHg; Level 3 adds HOMA-IR 1.5, TG 100 mg/dL and HDL 50 mg/dL.

AI guidance

The server recomputes every score from your numbers and gives Claude only those results, so the AI cannot change or contradict them. It is told not to diagnose, not to name medicines or doses, and to put urgent flags first. Its output is checked against a strict format before it is shown. If the AI is unavailable, rate-limited or declines, the rule-based plan is shown instead.

Privacy

Calculations run in your browser and your entries are saved only on this device (and in the link if you copy it). When you press Get AI guidance, the numeric values and answers for the current level are sent to this site's server and on to Anthropic's API to write the plan. No name, email or free text is included in that request. Nothing else is sent anywhere unless you choose to: if you tick the consent box and press Send my results, your optional name and email, the values you entered for the current level and your result are sent to the site owner through Formspree (formspree.io), which stores them for the owner. That is health information, so only send it if you are comfortable doing so. Ask the site owner to delete it at any time.

Limitations

• No result here is a diagnosis. The questionnaires were validated in Finnish (FINDRISC) and Indian (IDRS) populations and may be less accurate elsewhere.
• HbA1c can be misleading with anaemia, haemoglobin variants, pregnancy or recent blood loss.
• FIB-4 is less reliable under 35 and over 65. eGFR from creatinine is affected by muscle mass and diet.
• Medications that change glucose, lipids or blood pressure are not modelled.

References

1. Lindström J, Tuomilehto J. The Diabetes Risk Score: a practical tool to predict type 2 diabetes risk. Diabetes Care 2003;26:725–731.
2. Mohan V, Deepa R, Deepa M, Somannavar S, Datta M. A simplified Indian Diabetes Risk Score for screening for undiagnosed diabetic subjects. J Assoc Physicians India 2005;53:759–763. TODO review: activity item mapping
3. American Diabetes Association Professional Practice Committee. Diagnosis and Classification of Diabetes: Standards of Care in Diabetes. Diabetes Care, current Supplement 1. TODO: pin edition & pages
4. Alberti KGMM et al. Harmonizing the metabolic syndrome. Circulation 2009;120:1640–1645.
5. WHO Western Pacific Region / IASO / IOTF. The Asia-Pacific perspective: redefining obesity and its treatment. 2000.
6. Matthews DR et al. Homeostasis model assessment. Diabetologia 1985;28:412–419.
7. Sterling RK et al. Development of a simple noninvasive index to predict significant fibrosis (FIB-4). Hepatology 2006;43:1317–1325.
8. Inker LA et al. New creatinine- and cystatin C–based equations to estimate GFR without race. N Engl J Med 2021;385:1737–1749.
9. Diabetes Prevention Program Research Group. Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin. N Engl J Med 2002;346:393–403.
10. Whelton PK et al. 2017 ACC/AHA high blood pressure guideline. Hypertension 2018;71:e13–e115.