Multi-modal infusion pump real-time monitoring technique for improvement in safety of intravenous-administration patients

Abstract

Intravenous (IV) medication administration processes have been considered as high-risk steps, because accidents during IV administration can lead to serious adverse effects, which can deteriorate the therapeutic effect or threaten the patient’s life. In this study, we propose a multi-modal infusion pump (IP) monitoring technique, which can detect mismatches between the IP setting and actual infusion state and between the IP setting and doctor’s prescription in real time using a thin membrane potentiometer and convolutional-neural-network-based deep learning technique. During performance evaluation, the percentage errors between the reference infusion rate (IR) and average estimated IR were in the range of 0.50–2.55%, while those between the average actual IR and average estimated IR were in the range of 0.22–2.90%. In addition, the training, validation, and test accuracies of the implemented deep learning model after training were 98.3%, 97.7%, and 98.5%, respectively. The training and validation losses were 0.33 and 0.36, respectively. According to these experimental results, the proposed technique could provide improved protection functions to IV-administration patients.

Convolutional neural network-based ambient light-independent panel digit surveillance technique for infusion pumps

“…Furthermore, the accuracies of the model were 98.3%, 97.7%, and 98.5% for training, validation, and test dataset in our previous study. 5 In the current study, the accuracies of the model in day-lighting mode were 99.9% for both training and validation dataset, and 98.2% during continuous 24-h monitoring test, which demonstrated almost even performance in day-lighting environments. Utilizing this result, for example, long-term, continuous closed-loop monitoring of infusion pump operation can be realized by combining the currently proposed ambient light-independent panel digit monitoring technique and the previously proposed membrane sensor-based actual infusion rate monitoring technique.…”

“…Utilizing this result, for example, long-term, continuous closed-loop monitoring of infusion pump operation can be realized by combining the currently proposed ambient light-independent panel digit monitoring technique and the previously proposed membrane sensor-based actual infusion rate monitoring technique. 5,17…”

“…In a previous study, 5 the authors proposed a surveillance technique involving a convolutional neural network (CNN) that can monitor variations in the panel digits of an infusion pump in real-time using a camera module fixed in front of the pump panel. In that study, a pattern marker was attached to a pre-determined position within the front panel to extract the ROI positions for the rate and volume panel.…”

Deep learning-based monitoring technique for real-time intravenous medication bag status

“…However, the currently proposed smart IV pole can only monitor variations in IV bag status and cannot monitor the inaccurate setting/operation of electronic IV infusion pumps that are connected to the IV bags to control the infusion rate and infusion volume of medications in each of the IV bags for critically ill patients. To further expand the role of the IV pole in monitoring pump-related IV accidents, it is necessary to implement additional DL models that can extract the state of panel settings of various electronic IV infusion pumps and automatically determine the relevance of the device setting in further studies [ 28 ]. In addition, the electronic syringe pump uses a drug-filled syringe instead of an IV bag; therefore, more dedicated techniques for infusion rate and drug residue monitoring from the pump-equipped syringe, as well as pump-setting monitoring using DL models, are required.…”

Development of smart infusion pumps: State of the art and future perspectives

“…It can achieve more accurate monitoring of low-flow intravenous infusion. Hwang et al 41 proposed a multi-modal IP monitoring technique. Using a thin membrane potentiometer and convolutional-neural-network–based deep learning technique, it can detect mismatches between the IP setting and actual infusion state and between the IP setting and doctor’s prescription in real-time.…”