Dialysis machine alarm recognition based on convolutional neural network

Abstract Background Hemodialysis, a renal replacement treatment for end-stage renal failure, relies heavily on the proper functioning of the dialysis machine. Timely detection and handling of dialysis machine alarms are important to ensure the safety of dialysis treatment. Method This study proposes...

Full description

Saved in:
Bibliographic Details
Main Authors: Huile Xie, Xiongjie Deng, Bin Dong, Liting Chen, Mingyang Song, Zidong Ying, Zhaoxin Fan, Xukai Wang, Liang Peng
Format: Article
Language:English
Published: BMC 2025-06-01
Series:Renal Replacement Therapy
Subjects:
Online Access:https://doi.org/10.1186/s41100-025-00632-9
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Abstract Background Hemodialysis, a renal replacement treatment for end-stage renal failure, relies heavily on the proper functioning of the dialysis machine. Timely detection and handling of dialysis machine alarms are important to ensure the safety of dialysis treatment. Method This study proposes a method for recognizing dialysis machine alarms using a convolutional neural network (CNN). A dataset of dialysis machine alarm light images was created through a multicenter collaboration, which was used to train the YOLOv5 model. Results The study shows that the average recognition precision, recall, and mAP@0.5 for each warning light category reached 0.892, 0.813, and 0.833, respectively. A well-trained model can quickly and accurately recognize a variety of dialysis machine alarm types. Conclusions It is feasible to use convolutional neural networks to recognize dialysis machine alarms, and they can be widely used to improve dialysis safety and management.
ISSN:2059-1381