Diabetes Prediction API

The Diabetes Prediction API is a machine learning-based API designed to predict whether a person is diabetic based on various health metrics. The model used is a Support Vector Machine (SVM) with an accuracy of 77.27%. The API is developed using FastAPI and deployed on Render.

Deployed Model

• API Endpoint: https://dibetes-api.onrender.com/diabetes_prediction

Usage

Input

The API expects a JSON object with the following fields:

• Pregnancies: Number of times pregnant (integer)
• Glucose: Plasma glucose concentration (integer)
• BloodPressure: Diastolic blood pressure (integer)
• SkinThickness: Triceps skin fold thickness (integer)
• Insulin: 2-Hour serum insulin (integer)
• BMI: Body mass index (float)
• DiabetesPedigreeFunction: Diabetes pedigree function (float)
• Age: Age in years (integer)

Output

The API will return a JSON object with the prediction:

• "The Person is Diabetic" if the person is predicted to be diabetic.
• "The Person is not Diabetic" if the person is predicted to be non-diabetic.

Example Requests

Test Case 1

Input:

{
    "Pregnancies": 10,
    "Glucose": 168,
    "BloodPressure": 74,
    "SkinThickness": 0,
    "Insulin": 0,
    "BMI": 38,
    "DiabetesPedigreeFuncion": 0.537,
    "Age": 30
}

Result:

"The Person is Diabetic"

Test Case 2

Input:

{
    "Pregnancies": 1,
    "Glucose": 85,
    "BloodPressure": 66,
    "SkinThickness": 29,
    "Insulin": 0,
    "BMI": 26.6,
    "DiabetesPedigreeFuncion": 0.351,
    "Age": 31
}

Result:

"The Person is not Diabetic"

API Testing

You can test the API using tools like Postman. Here is how you can do it:

1. Open Postman.
2. Create a new POST request.
3. Set the URL to https://dibetes-api.onrender.com/diabetes_prediction.
4. In the Body tab, select raw and JSON format.
5. Enter the JSON object as per the input format.
6. Send the request and check the response.

Conclusion

This Diabetes Prediction API provides an easy and efficient way to predict diabetes using health metrics. With an accuracy of 77.27%, it can be a valuable tool in preliminary health assessments.