Extended Version for the Diabetes Prediction : Please check the diabetes prediction first.
A diabetes prediction web application built using machine learning provides a simple and effective tool for early health assessment.
This project allows users to input basic medical parameters and instantly receive a prediction about whether the person may have diabetes.
Using Python and Streamlit the system delivers a lightweight interactive web interface for real time predictions making it helpful for awareness and preventive decision making.
Project Overview
- Machine learning powered diabetes prediction system
- Integrated into a Streamlit web app for user friendly interaction
- Uses a pre trained model loaded from a pickle file
- Accepts eight medical input values and returns a prediction result
Libraries Used
- NumPy for numerical array handling
- Pickle for loading the pre trained ML model
- Streamlit for building the web application interface
- Scikit Learn model used during offline training before export as a pickle file
Application Details
The project uses a previously trained machine learning model saved as
trained_model.sav
The app accepts the following health features from the user
- Number of pregnancies
- Glucose level
- Blood pressure
- Skin thickness
- Insulin level
- BMI
- Diabetes pedigree function
- Age
These inputs are fed into the model to determine whether the person is diabetic or not.
Preprocessing Steps
- User values entered through text input fields in Streamlit
- Inputs are collected as a list and converted to a NumPy array
- Array is reshaped to match the model input size so the model understands it is a single person prediction
- Model produces a binary output where
- Zero means not diabetic
- One means diabetic
Model Interaction
- The web app loads the trained model using pickle load
- Function diabetes prediction handles
- Input conversion
- Reshaping
- Model prediction
- Returning a readable output message for the user
Example logic from your app
- If prediction equals zero return person is not diabetic
- Otherwise return person is diabetic
Streamlit Interface
- App title displayed using st.title
- Text input fields collect user data from browser
- When the user clicks the Diabetes Test Result button the result is calculated
- st.success displays the output message on screen
The interface is simple clean and suitable for both beginners and health analysts.
Prediction Flow
1 User enters eight medical values
2 Values are converted into numerical format
3 Model predicts the health status
4 Result is displayed immediately on the Streamlit page
Deployment Possibilities
- Deploy on Streamlit Cloud for instant public access
- Package inside a Flask app for full scale web deployment
- Embed inside healthcare portals or community screening tools
Key Takeaways
- Complete working ML web app built using Streamlit
- Demonstrates how offline trained models can be deployed through a simple UI
- Shows practical application of machine learning for public health awareness
Future Enhancements
- Add validation to check if user inputs are numeric
- Improve UI with sliders instead of text fields
- Add data visualizations for health insights
- Deploy as a mobile friendly interface