FL4Hospital is my bachelor thesis at the University of Stavanger, completed in collaboration with two other students. The project explores how federated learning, IoT devices, and smart-room automation can improve patient comfort while keeping patient data private.
The system combines a synthetic hospital simulation with a hybrid prototype. BLE-connected Nordic sensors collect environmental data, while patient preferences, medication, nurse visits, and room activity inform comfort predictions and automated room adjustments. Separate fixed-time and event-driven federated learning pipelines train models across room-based datasets without centralizing sensitive data.
Python powers the simulation, federated learning pipelines, BLE room runtime, dashboards, and SQLite/SQLAlchemy persistence layer. The project also includes HTML-based dashboard interfaces and real-time environmental sensing with Nordic devices.