Predict Diabetes
Before It Starts Powered by an advanced multi-model AI ensemble

D-Forecast AI uses a powerful ensemble of specialised AI models trained on diverse clinical populations to generate a personalised diabetes risk score — accurate, explainable, and actionable.

Our mission
Save lives early
Catch risk before symptoms appear — when intervention is cheapest and most effective
Democratise screening
Free, instant access to hospital-grade risk intelligence — no doctor visit required
Empower with knowledge
Explain every score in plain language — because understanding drives action
Bridge the gap
Connect individuals to the right clinical care at the right moment — not too late
99%
Prediction accuracy
4+
Risk tiers classified
12+
Health signals analysed
🧬
537M
Adults living with diabetes globally
50%
Undiagnosed — unaware of their condition
80%
Type 2 cases preventable with early action
$966B
Annual global healthcare expenditure
Process

Three steps to your
risk intelligence report

No lab required for a first estimate. Optional biomarker inputs allow the AI to go deeper for higher accuracy.

01
📋

Enter Your Health Profile

Provide basic clinical values — age, BMI, fasting glucose, HbA1c. Optionally add lifestyle factors and lipid panel for a deeper, more accurate assessment.

02

Ensemble Models Compute

D-Forecast's AI engine dynamically activates the right analytical models for your input data. More data provided means more signals cross-validated for a higher confidence score.

03
🎯

Receive Actionable Report

Get a calibrated risk percentage, tier classification (Low → Very High), and personalised, actionable recommendations tailored to your specific health profile.

Smarter than any
single algorithm.

D-Forecast's AI combines multiple specialised models trained across diverse clinical populations — each contributing unique insight to your final risk score.

All systems operational
Core Clinical
Clinical Risk Analysis
Analyses your fundamental clinical markers — blood glucose, HbA1c, BMI and age — using models trained on large community population datasets. Always active.
InputsGlucose, HbA1c, BMI, Age
AvailabilityAlways active
Confidence levelHigh
Lifestyle Aware
Lifestyle Risk Engine
Accounts for behavioural and environmental risk factors. Activates automatically when lifestyle information is provided, increasing prediction depth significantly.
InputsSmoking, Hypertension, Activity
AvailabilityActivates with lifestyle data
Confidence boost+20%
Biomarker Deep Dive
Lipid Biomarker Analysis
Uses lipid panel values to perform a 3-class classification: Non-Diabetic, Pre-Diabetic, or Diabetic. The only layer that explicitly detects the pre-diabetic window.
InputsHDL, Triglycerides, LDL
AvailabilityActivates with lipid data
Unique capabilityPre-diabetes detection
Population Coverage
Multi-Population Training
Our AI was trained across multiple diverse clinical populations — community surveys, hospital records, and metabolic cohorts — reducing demographic bias in predictions.
CoverageMultiple ethnicities & regions
PurposeReduce prediction bias
Population diversityHigh
Dynamic Weighting
Adaptive Ensemble Logic
Each AI layer's contribution is dynamically weighted based on the data you provide. More complete inputs shift weight toward more specialised models for higher accuracy.
MechanismContext-aware weighting
EffectHigher confidence with more data
AdaptabilityDynamic
Early Detection
Pre-Diabetic Signal Boost
A calibrated nudge layer that amplifies early warning signals — borderline HbA1c, slightly elevated glucose, and BMI thresholds — to catch risk before it becomes diagnosis.
FocusBorderline & subclinical risk
GoalCatch it before it's too late
Early catch rateHigh

How your score
is calculated.

Your final risk score blends three independent signal layers — more data you provide, more layers activate, and the higher the confidence.

Clinical signalsCore
Metabolic profileCore
Glucose & HbA1c patternCore
Lifestyle risk layer+Boost
Lipid biomarker layer+Boost
Early-detection nudge+Fine-tune
Weighted blend
∑wᵢpᵢ
+ Lifestyle score
+ Biomarker heuristic
+ Early-detection nudge
Risk tier output
● Low <15%
● Moderate 15–35%
● High 35–60%
● Very High >60%

Diabetes is silent.
Until it isn't.

Most Type 2 diabetes develops over years with few symptoms. By the time clinical diagnosis happens, significant damage may already be done. Early risk prediction changes everything.

🔬

Pre-diabetes is reversible

With early lifestyle intervention, pre-diabetic individuals can reduce progression risk by up to 58%. D-Forecast is designed to catch risk in this critical, reversible window.

📊

Multi-signal analysis reduces errors

A single data point can be misleading. D-Forecast cross-validates across clinical, lifestyle, and biomarker signals, significantly reducing the chance of a missed high-risk case.

💊

Early action saves lives and cost

Treating diabetes costs 2–3× more than prevention. Screening and early intervention programmes show consistent 30–40% cost reduction over 10 years.

Interactive risk estimator
28%
Moderate Risk
Low
Moderate
High
Very High
40
26
100 mg/dL
5.5%

⚠ This illustrative estimator uses simplified heuristics only. Use the full prediction tool for your actual risk assessment.

Education

Understanding the
types of diabetes

Type 2
Type 1
Pre-Diabetes
Gestational

Type 2 Diabetes

The most common form, accounting for over 90% of all cases. Develops when cells become resistant to insulin or the pancreas doesn't produce enough. Strongly linked to lifestyle factors.

Unlike Type 1, it is preventable and often reversible in early stages with diet, exercise and weight management. D-Forecast is primarily optimised for Type 2 prediction.

Onset typically gradual over months or years
Risk increases with age, obesity, inactivity
HbA1c ≥ 6.5% is diagnostic threshold
Early lifestyle changes can prevent or delay onset
90%
Of all diabetes
~462M
People affected
58%
Preventable
6.5%
HbA1c threshold

Type 1 Diabetes

An autoimmune condition where the immune system mistakenly destroys insulin-producing beta cells in the pancreas. Requires lifelong insulin therapy.

Not preventable and not associated with lifestyle factors. Our models do not screen for Type 1 specifically, though high HbA1c and glucose values will still flag as high-risk.

Autoimmune destruction of beta cells
Onset typically in childhood or young adults
Requires insulin therapy for life
Not preventable — unrelated to lifestyle
8–10%
Of all diabetes
~8.4M
People globally
0%
Preventable
Any
Age of onset

Pre-Diabetes

Blood glucose levels higher than normal but not yet high enough for a Type 2 diagnosis. A critical warning window — intervention here is most effective.

The Iraqi Biomarker model in our ensemble specifically identifies pre-diabetic metabolic patterns via lipid profiles. Many people with pre-diabetes don't know they have it.

Fasting glucose 100–125 mg/dL
HbA1c 5.7–6.4% is pre-diabetic range
Often no symptoms — detected by testing
58% risk reduction with lifestyle changes
374M
With pre-diabetes
70%
Will develop T2DM
5.7%
HbA1c threshold
Fully
Reversible

Gestational Diabetes

Occurs during pregnancy when blood glucose levels rise above normal. Usually resolves after birth but significantly increases lifetime risk of Type 2 diabetes for both mother and child.

D-Forecast AI is designed to support general diabetes risk screening and may be used alongside clinical judgement for maternal risk assessment.

Develops during 2nd or 3rd trimester
Usually resolves after delivery
Increases T2DM risk by 7× later in life
Managed through diet, exercise, sometimes insulin
21M
Births affected
Later T2DM risk
2–10%
Of pregnancies
Usually
Temporary
Get Started

Know your risk.
Take control.

Run a full AI-powered prediction in under 2 minutes. No account needed. Results are private and stored locally in your session.

Clinical Analysis
Lifestyle Scoring
Lipid Biomarkers
Pre-diabetes Detection
Personalised Recs
Start Free Assessment