A Data Science & Machine Learning Project integrating GRU-based forecasting and HUOMIL pattern mining to optimize hospital operations.
HASTURE is an intelligent Hospital Management System (HMS) built with Django, GRU deep learning, and HUOMIL pattern mining.
It addresses critical hospital challenges:
- 🏷️ Automating operations: Patient admission, ward allocation, inventory management
- 📊 Forecasting inventory: GRU-based time series model for 12-month predictions
- 🧮 Pattern mining: HUOMIL algorithm to identify high-utility & high-occupancy medical supplies
- 📈 Dashboards: Role-based dashboards for doctors, in-charges, inventory managers, and administrators
- 🔒 Security: Role-based authentication & access control
- Languages: Python (NumPy, Pandas, Scikit-learn, TensorFlow/Keras)
- Frameworks: Django, MySQL
- Machine Learning: GRU (Time Series Forecasting), HUOMIL (Pattern Mining)
- Visualization: Matplotlib, Seaborn, Plotly
- Tools: Jupyter Notebook, Excel (openpyxl), GitHub
- GRU Forecasting – Predicts 12-month demand for hospital inventory with MSE ≈ 9.74
- HUOMIL Algorithm – Mines high-utility, frequently used medicines (e.g., Paracetamol, Amoxicillin, Insulin)
- Dashboards – Real-time ward occupancy, patient trends, and inventory usage
- Role-based Access – Secure, tailored dashboards for Admin, Doctor, In-Charge, and Inventory Manager
- Transformer-based forecasting models
- Adaptive HUOMIL thresholds
- Real-time anomaly detection
- Mobile dashboard access
- NLP-based chatbot queries
- ✅ Reduced inventory wastage via GRU predictions
- ✅ Targeted procurement with HUOMIL insights
- ✅ Improved patient care through automated workflows
- ✅ Smarter hospital decisions with role-based dashboards