AI-Powered Drug Interaction Alert System | Machine Learning Healthcare Final Year Project with Source Code

The Drug Interaction Alert System is a machine learning powered web application that helps you check whether two medicines are safe to take together. It searches a real dataset of 191,541 drug interactions sourced from DrugBank, and instantly tells you how risky a combination is by labelling it as Major, Moderate, or Minor. If a drug pair is not present in the database, a trained Random Forest model steps in and predicts the severity on its own, so the system always returns a useful answer.

This project is built for students who want something more meaningful than a basic CRUD app. It touches real healthcare data, a complete machine learning pipeline, and a clean web interface, which makes it a strong choice among AI and machine learning final year projects for BCA, MCA, B.Tech CSE, and B.Sc IT students.

Key Features

• Instant interaction lookup across 191,541 known drug pairs from DrugBank.
• Machine learning fallback using a Random Forest model that predicts severity when a pair is missing from the database.
• Live autocomplete with type-ahead search across 1,701 drugs and full keyboard navigation.
• Severity classification shown as colour coded Major, Moderate, and Minor badges with short clinical guidance.
• Search history that keeps your last ten lookups within the session.
• JSON API so the system can be plugged into other applications.
• Responsive interface that works smoothly on desktop, tablet, and mobile, with no external JavaScript frameworks.

How It Works

Every search first runs against the CSV lookup table that holds the 191K known interactions. If the pair is found, the app returns the original interaction description along with a rule based severity level. If the pair is not found, the request is passed to the machine learning model, which uses TF-IDF features built from interaction descriptions along with encoded drug names to predict the severity. The result page clearly marks whether the answer came from a Database Match or an ML Prediction, so the output stays transparent.

Machine Learning Pipeline

The included Jupyter notebook walks through the entire workflow. Drug names are cleaned and lowercased, interaction descriptions are converted into 5,000 TF-IDF features, and drug names are label encoded into a combined sparse feature matrix. Three classifiers are trained on an eighty twenty stratified split, namely Random Forest, XGBoost, and Logistic Regression, and the best performing model is saved as a pickle file for the web app to load. The notebook also produces clear visualisations such as the top interacting drugs, an interaction type distribution, a confusion matrix, and a drug interaction network graph.

Real World Applications

• A learning aid for pharmacy and medical students to understand interaction severity.
• A reference tool that small clinics or labs can adapt for quick internal checks.
• A base project for building larger healthcare decision support systems.
• A clear demonstration of how rule based logic and machine learning can work together in one product.

Why This Makes a Strong Final Year Project

This project covers three areas that examiners genuinely value, namely real data handling, applied machine learning, and full stack web development. It is easy to demo live, the results are explainable, and the healthcare angle gives your viva discussion real depth. If you like this kind of data driven build, you may also want to look at our Traffic Accident Severity Prediction project, which uses a similar Flask and machine learning approach.

Technology Stack

The system uses Python and Flask on the backend, scikit-learn and XGBoost for the models, TF-IDF for feature engineering, and pandas with NumPy for data processing. The frontend is built with clean HTML5, CSS3, and vanilla JavaScript, so there are no heavy framework dependencies to manage during setup.

What You Get

You receive the complete source code, the trained model files, the dataset, and the training notebook. You can buy it directly and start running it the same day. To explore more options, browse all our final year projects with source code.

Add-On Services

• Idea Implementation: tell us your own concept and we build a custom project for you.
• Project Setup and Source Code Explanation: a live session where we install the project and walk you through the full logic and flow. Learn more on our project setup and explanation page.
• Custom Report, Research Paper, and PPT: we prepare college format documentation and can guide you through research paper writing and publishing.

Note: This project is intended for educational and research purposes. The interaction data is sourced from DrugBank. Always consult a qualified healthcare professional before making any medical decision.