DerZiad/AI-Based-Diabetic-Patient-Classification

This project was developed as part of my fifth-semester Artificial Intelligence coursework in Morocco. The goal is to design and compare machine learning and deep learning models capable of classifying patients as diabetic or non-diabetic. # 🩺 Diabetic Patients Classification Project This project was developed as part of my fifth-semester Artificial Intelligence coursework in Morocco. The goal is to design and compare **machine learning** and **deep learning** models capable of classifying patients as *diabetic* or *non-diabetic*. The challenge lies in identifying the most effective model — balancing accuracy, generalization, and computational efficiency — to support informed healthcare decision-making. --- ## 🚀 Project Overview The application provides a **user-friendly interface** for both data input and prediction: * Users can **manually enter patient data** or **import data from an Excel file**. * The system processes the data and predicts the likelihood of diabetes using pre-trained models. * The predictions are powered by two optimized models, chosen for their high accuracy and ability to reduce overfitting. --- ## 🧠 Technologies Used The project integrates several key technologies and libraries: | Purpose | Technology | | ----------------- | -------------------------- | | User Interface | **PyQt5** | | Deep Learning | **TensorFlow** | | Machine Learning | **Scikit-learn (Sklearn)** | | Data Processing | **Pandas** | | Model Persistence | **Pickle** | | CLI Enhancements | **Colorama** | --- ## 🧩 Models Implemented The classification system leverages multiple models to ensure robust performance: ### **Machine Learning Models** * Random Forest Classifier * Support Vector Machine (SVM) * Logistic Regression * K-Nearest Neighbors (KNN) ### **Deep Learning Model** * Fully connected (Dense) neural network built with TensorFlow. --- ## ⚙️ Optimization Techniques To maximize model performance and reliability, the following strategies were applied: * **Data Scaling** – Normalizing data to ensure uniform feature contribution. * **K-Fold Cross-Validation** – Providing robust …