Embedded Files
Summary of Findings
Summary of Findings
- Dataset: 768 female patients, 8 health features
- Key predictors: Glucose, Insulin, BMI
Random Forest selected as best model:
Random Forest selected as best model:
- Accuracy: 88.67%
- Recall: 0.922 (best for catching true diabetic cases)
- Strong F1 score and AUC
Clinical Relevance of the Model
Clinical Relevance of the Model
- Identifies critical health markers for early intervention
- Can assist doctors in risk screening and prioritization
- Deployable in health apps or clinic decision systems
Limitations & Future Improvements
Limitations & Future Improvements
Limitations:
Limitations:
- Small dataset which only includes female patients
- Missing behavioral/lifestyle data
- Needs validation on diverse populations
Further enhancement :
Further enhancement :
- Add interpretable ML (e.g., SHAP, LIME)
- Explore deep learning for richer feature patterns
OUR PROJECT MATERIALS:
OUR PROJECT MATERIALS:
CLASSIFICATION NOTEBOOK:
CLASSIFICATION NOTEBOOK:
[https://colab.research.google.com/drive/1XHUWHskALSlictuN7_NW50nronkajl27?usp=sharing]
[https://colab.research.google.com/drive/1XHUWHskALSlictuN7_NW50nronkajl27?usp=sharing]
CLUSTERING NOTEBOOK:
CLUSTERING NOTEBOOK:
[https://colab.research.google.com/drive/1mK8jDv7Bllib64lZBgYuj-jF4icvISwH?usp=sharing]
[https://colab.research.google.com/drive/1mK8jDv7Bllib64lZBgYuj-jF4icvISwH?usp=sharing]
USER TESTING WEBSITE:
USER TESTING WEBSITE:
[https://diabetes-prediction-exkn.onrender.com/#]
[https://diabetes-prediction-exkn.onrender.com/#]
Page updated
Google Sites
Report abuse