← Back to Portfolio
Diabetes Risk Prediction Engine
Project 05 ✓ Peer Reviewed
Patient Clinical Input
Enter patient clinical values to receive an instant diabetes risk prediction powered by the published gradient boosting model (82%+ accuracy).
Glucose Level (mg/dL) Normal: 70–100
148
BMI (kg/m²) Normal: 18.5–24.9
33.6
Age Years
50
Blood Pressure (mmHg) Diastolic
72
Insulin Level (IU/mL) Fasting normal: <25
Skin Thickness (mm)
Diabetes Pedigree Function 0.0 – 2.5
0.627
Pregnancies Number of times
Prediction Result
72%
HIGH RISK
Clinical indicators suggest elevated probability of diabetes. Immediate medical consultation recommended.
📊 Feature Contribution to Risk
Glucose Level (148 mg/dL)Very High Impact
Diabetes Pedigree (0.627)High Impact
BMI (33.6)High Impact
Age (50)Moderate Impact
Blood Pressure (72 mmHg)Low Impact
💬 Clinical Explanation
The primary risk driver is a glucose level of 148 mg/dL, which is significantly above the normal fasting range of 70–100 mg/dL and enters the pre-diabetic threshold (≥126 mg/dL). Combined with a BMI of 33.6 (classified as obese) and a diabetes pedigree function of 0.627 indicating strong family history, this profile closely matches positive cases in the training dataset. Age (50) further elevates risk as the model identified age as a moderate predictor beyond 45 years.
✅ Recommendations
🏥
Consult a physician immediately for HbA1c testing to confirm or rule out diabetes diagnosis.
🥗
Dietary changes — reduce refined carbohydrate intake and consider a low-glycaemic diet plan.
🏃
Increase physical activity — 30 minutes of moderate exercise 5 days per week significantly reduces glucose levels.
📊
Monitor glucose weekly and track trend over 8 weeks before next clinical review.
🔬 Model Performance Metrics
82%
Test Accuracy
78%
Recall (Positive)
0.84
ROC-AUC Score
768
Training Records