Development of a Smartphone-Based Inventory Management System for Emergency Carts

Area of Science:

• Medical Informatics
• Computer Vision
• Healthcare Management

Background:

• Hospital inventory management for emergency carts is a time-consuming task for healthcare staff.
• High workload and pressure on healthcare professionals necessitate streamlined workflows.

Purpose of the Study:

• To develop and evaluate an automated inventory management system for emergency carts using computer vision.
• To assess the impact of the system on task completion time and healthcare staff workload.
• To compare the performance of various object detection algorithms for automatic drug counting.

Main Methods:

• Development of a smartphone-based computer vision system for emergency cart inventory.
• User study involving 37 nurses to compare task time before and after system implementation.
• Evaluation of YOLOv4, YOLOv5, YOLOv7, and YOLOv5-OBB object detection algorithms for drug counting using 500 training images.

Main Results:

• Average inventory task time decreased from 9.59 minutes to 5.32 minutes.
• Average nurse workload score reduced from 3.81 to 2.70.
• A hybrid approach combining YOLOv5-OBB and YOLOv7 achieved the highest accuracy in drug counting, with 1.29% over-detection and 3.27% missed detection rates.

Conclusions:

• Computer vision technology, particularly YOLO object detectors, can effectively automate emergency cart inventory management.
• The proposed system streamlines workflows and significantly reduces the workload for healthcare personnel.
• Smartphone-based solutions offer a practical approach to improving hospital inventory management efficiency.