Smart MRI Departments: Intelligent Patient Scheduling Using Real-Time Appointment Duration Metrics and an FNN to Predict Lateness

Motivation: A significant challenge in MRI departments is accurately scheduling patient appointments to maximise efficiency while minimising patient backlog and delays. Goal(s): To develop an AI-driven tool that uses patient demographic data, scanner utilisation metrics (for real-time appointment duration), and RFID information to create a comprehensive patient booking system. Approach: 40,000 sets of patient demographic data were processed to train an FNN-based network that estimates patient punctuality. Scanner log files were analysed to ascertain MRI metrics such as accurate protocol lengths. Results: The FNN outperforms current techniques when predicting patient arrival times and we can extract accurate protocol times for training our next model. Impact: By increasing patient throughput, our work will be a crucial step towards a smart, sustainable MRI department. More patients will be scanned with less scan idle-time, improving patient outcomes in the long-term, whilst contributing to a greener and cost-effective service.